Molecular Insights on the Solvent Effect of Methanol
Additive in Glycine Polymorph Selection
by
Srikanth Patala
Submitted to the Department of Materials Science and Engineering
in partial fulfillment of the requirements for the degree of
Master of Science
MASSACHUSETT•
INSTITUTE
OF TECHNOLOGY
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUN 16 2008
May 2008
LIBRARIES
@ Massachusetts Institute of Technology 2008. All rights reserved.
.kRCHIVES
Author ......
Department of Materials Science and Engineering
May 30, 2008
Certified by.
..
. . . . .. . . . . ..
Bernhardt L. Trout
Associate Professor of Chemical Engineering
Thesis Supervisor
Certified by............
Samuel M. Allen
POSCO Professor of Physical Metallurgy
Thesis Reader
Accepted by .........
Samuel M. Allen
POSCO Professor of Physical Metallurgy
Chair, Departmental Committee on Graduate Students
Molecular Insights on the Solvent Effect of Methanol Additive
in Glycine Polymorph Selection
by
Srikanth Patala
Submitted to the Department of Materials Science and Engineering
on May 30, 2008, in partial fulfillment of the
requirements for the degree of
Master of Science
Abstract
In an effort to improve control and design in organic crystallization, the effect of solvent on polymorph selection has gained tremendous interest in recent years. In this
thesis, molecular simulation techniques are used to gain insight into the solvent effect
on glycine crystallization in water-methanol mixtures. We report the validation of the
Optimized Potential for Liquid Simulations (OPLS) force field and parameters with
modified Lennard-Jones parameters for hydrogens attached to a-carbon in glycine
zwitterion. Solution and interface simulations in water and 50% v/v water-methanol
solutions reveal the mechanism through which methanol additive results in the crystallization of the least stable P-glycine polymorph. Free energy calculations through
the Umbrella Sampling method show an increased stability of the centrosymmetric
dimer structure (a-glycine growth unit) in the presence of the methanol additive.
Even though the dimer structure is more stable in water-methanol mixtures, a higher
fraction of glycine monomers were observed in water-methanol mixtures. It is revealed through thermodynamic arguments that a drastic decrease in solubility results
in a higher fraction of glycine monomers in water-methanol mixtures. It was hypothesized in previous studies that the presence of monomer units docking onto the (010)
interface of a-glycine inhibits further growth due to exposed ammonium groups at the
interface. The effect of solvent on crystal growth inhibition is explored by the interface simulations of a-glycine in water-methanol mixtures. When the monomer units
are docked onto the interface, water is shown to be more effective than methanol
in inhibiting crystal growth of (010) interface of a-glycine. This study sheds light
on the role played by the solvent on glycine polymorph selection in water-methanol
solutions.
Thesis Supervisor: Bernhardt L. Trout
Title: Associate Professor of Chemical Engineering
Acknowledgments
My experience at MIT has been both exciting and challenging to say the least. The
most exciting part is getting the opportunity to work on this project under the guidance of my advisor, Prof. Bernhardt L. Trout. I am extremely grateful for his belief
in me and also for his constant support and invaluable advice throughout the project.
I owe a great deal to Dr. Gregg Beckham without whom this study would be
incomplete. I thank him for teaching me how to use CHARMM and also for many
invigorating and creative ideas. I am greatly indebted to Jie Chen for her assistance
during many stages of this project, Dr. Naresh for his encouragement and advice
throughout, and Bin Pan for his help in solving many technical problems and for his
wonderful Chinese tidbits. The Trout Group presented me with a stimulating and
fun filled atmosphere and I am very thankful for the opportunity to interact with
them. A special thanks to Prof. Samuel Allen, Materials Science Department Chair,
who has been very kind and helpful in sharing his invaluable time with me during the
course of my stay here.
Any of my experiences at MIT would not have been so compelling without some
of the wonderful people that I have gotten to know in the last two years: Srujan,
with whom I had many memorable and inspirational moments; Dipanjan and Sukant
who endured all my tantrums as roommates and have been a great source of encouragement; Chetan, Ashish and Deep who always had insightful opinions about rmy
professional and personal life. I would also like to thank my parents and family who
have given me this great opportunity and always encouraged me to follow my dreams.
Finally, I would like to thank the Singapore MIT Alliance Flagship Research Program and the Department of Materials Science and Engineering for financial supp(ort
during my studies.
Contents
1 Introduction
1.1
Objectives & Overview ..........................
2 Organic Crystal Polymorphism
2.1
Introduction .......................
2.2 Nucleation Controlled Polymorph Selection ...............
2.3
2.2.1
Nucleation Theory . . . . . . . . . . . . . . ..........
2.2.2
Growth Synthon Hypothesis ...................
Growth Controlled Polymorph Selection
................
2.3.1
Crystal Growth in Solutions ...................
2.3.2
Crystal Growth in Polar Organic Crystals ("Relay Type" Growth
Mechanism) ...................
......
...
3 Glycine Polymorphism
33
3.1
Glycine and its Polymorphs
3.2
Solvent Effects on Glycine Polymorphism ...........
. . . . .
35
3.3
Glycine Polymorph Selections in Water-Methanol Solutions
. . . . .
37
3.3.1
Precipitation of /-glycine .
. . . . . 37
3.3.2
Inhibition of y-glycine
3.3.3
Inhibition of a-glycine ..................
..................
...............
.................
4 Validation of Force Field Parameters
. . . . . 33
. . . . . 38
. . . ..
40
5 Simulation Methods & Results
5.1
51
Solution Behavior of Glycine . ..................
....
5.1.1
Dimer Lifetime ..........
5.1.2
Free Energy Calculations . ..................
51
..............
.
..
51
53
5.2
Solution Simulations of Glycine .....................
55
5.3
Simulations of a & 3 Glycine Interfaces . ................
61
5.3.1
System Setup ...............
5.3.2
Density Profiles .........................
5.3.3
Scatter Plots & Energetics . ..................
6 Conclusions & Future Work
...........
61
63
.
67
75
List of Figures
20
2-1 Polymorphs of ROY. [1] .........................
2-2
The link between self-assembly in solution and its crystal structure. .
2-3
Two polymorphic forms of 2,6-dihydroxybenzoic acid: (a) the metastable
24
form 1 dimer structure, and (b) the stable form 2 catemeric structure [21. 25
2-4 An illustration of various processes involved in the crystal growth process from aqueous solutions [3].
28
.....................
2-5 Packing arrangement of y-glycine delineated by crystal faces, as viewed
down the b-axis. The capped faces at the +c end of the polar axis
expose NH3 + groups at their surface the opposite faces expose C02groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
2-6 Packing arrangement of y-glycine as viewed down the b-axis, depicting
the relay mechanism at (001) interface. The bound water molecules are
ejected from the pockets as glycine from solution approach the interface. 31
2-7 A schematic representation of "relay type" growth mechanism. ....
3-1 a-glycine as viewed down the a-axis.
31
Crystal consists of hydrogen
34
bonded bilayers ...............................
3-2 0-glycine crystal structure as viewed along the a-axis illustrating the
three-dimensional hydrogen bond network. . .............
3-3
.
34
y-glycine crystal structure.(a) viewed down c-axis, (b) viewed parallel
to c-axis. The crystal consists of polar helical hydrogen bond structure
35
3-4 Packing arrangement of 3-glycine. The (010) "azure" and (010) "pink"
surfaces, are exposed at the interface.[4] . .........
.......
38
3-5
Packing arrangement of 7-glycine showing the pockets of fast growing
(001) face that are poisoned by the adsorption of ethanol and methanol
molecules (shown as "balls and sticks").[4] . ...............
3-6
39
y-glycine crystals observed in several crystallizations from a 1:1 waterethanol solution.[4] ............................
40
3-7 Packing arrangements of a-glycine (a) exposing weak solvent binding
C-H bonds to the solution at (010) surface (azure) or (b) exposing
strong solvent-binding N-H bonds to the solution at the (010) surface
(pink).[4]....
........
....
............
........
41
3-8
a-glycine crystals grown from 9:1 water-ethanol.[4]
4-1
Glycine zwitterion with all the atoms labeled. The lennard-jones pa-
. .........
41
rameters of H4 & H5 were modified to obtain a better agreement with
experimental observations ........
................
44
4-2 Radial distribution function calculated for the nitrogen atom in glycine
and oxygen atom in water .........................
4-3
47
Radial distribution function calculated for the hydrogen atom attached
to nitrogen in glycine and oxygen atom in water..............
48
4-4 Radial distribution function calculated for the oxygen atom in glycine
and oxygen atom in water ...........................
49
5-1
Dimer Structure of Glycine. .......................
52
5-2
Illustration of lifetimes of Dimer Structure in Water. The dark line at
3.75 A represents the average (dl + d2)/2 for the dimer structure. ..
5-3
52
Illustration of lifetimes of Dimer Structure in Water-Methanol mixtures. The dark line at 3.75
A represents
the average (dl + d2)/2 for
53
the dimer structure .............................
5-4 Free Energy diagram of dimerization in water and 50% v/v watermethanol solvents.
..................
.........
..
54
5-5 Fraction of glycine molecules that exist as monomers in glycine water
solutions at three different concentrations is shown. The thick lines
correspond to the average fraction of monomers. . ............
56
5-6 Fraction of glycine molecules that exist as monomers in glycine watermethanol solutions at three different concentrations is shown. The
thick lines correspond to the average fraction of monomers. .......
57
5-7 Fraction of glycine molecules that exist as monomers in water and
water-methanol mixtures at various supersaturations. . .........
60
5-8 Predicted fraction of glycine monomers in water and water-methanol
solvents and a hypothetical solvent at various supersaturations.
. ..
60
5-9 Fraction of glycine molecules that exist as dimers in the solution of
water and water-methanol solvents. . ...................
61
5-10 Schematic of the interface with water and (010) face of a-glycine with
C-H groups exposed. ..........................
62
5-11 Schematic of the interface with water-methanol mixture and (010) face
of a-glycine with N-H groups exposed . ................
63
5-12 Schematic of the interface with water-methanol mixture and (a) (010)
and (b) (010) of 0-glycine. ........................
63
5-13 Density of water on (010) and (010) interfaces of a-glycine with C-H
groups exposed at the interface. ....................
. .64 .
5-14 Density of water on (010) and (010) interfaces of a-glycine with N-H
groups exposed at the interface. . ..........
...
...
..
. .
65
5-15 Density of methanol on (010) and (010) interfaces of a-glycine with
N-H groups exposed at the interface.
. .................
65
5-16 Density comparison of solvent molecules on (010) interface of a-glycine
with C-H and N-H groups exposed at the interface.
. ........
66
5-17 Density of water on (010) and (010) interfaces of 0-glycine. .......
67
5-18 Density of methanol on (010) and (010) interfaces of a-glycine . . ..
67
5-19 Snapshot of the first structured layer on a-glycine interface with C-H
groups exposed ...........................
....
69
5-20 Snapshot of the first structured layer on a-glycine interface with N-H
groups exposed..............................
69
5-21 Center of Mass distribution of water molecules on the a-glycine interface with C-H groups exposed in the first structured layer. ......
70
5-22 Center of Mass distribution of water molecules on the a-glycine interface with N-H groups exposed in the first structured layer. ......
71
5-23 Center of Mass distribution of water molecules on the a-glycine interface with C-H groups exposed in the second structured layer. .....
72
5-24 Center of Mass distribution of water and methanol molecules on the
a-glycine interface with N-H groups exposed in the second structured
layer
..................................
.
73
List of Tables
4.1
Lennard Jones parameters for a-hydrogen . . ............
4.2
Polymorph Details for Glycine ......................
4.3
Number of Unit Cells and molecules per super cell during minimization
runs...................
4.4
.
45
.................
45
Root Mean Square Deviation (RMSD) of glycine polymorphs during
minimization using OPLS force field. . ..................
4.5
44
45
Percentage Deviation of lattice parameters of the three polymorphs of
glycine during a Inanosecond molecular dynamics simulation at 300 K
with OPLS force field ...........
4.6
......
..........
45
Relative lattice energies of glycine polymorphs obtained by annealing
the crystals to Zero Kelvin.
.......................
46
4.7 Average Coordination numbers with oxygen atoms in water. ......
4.8 Average Atom - Atom Distance ......
............
..
5.1
System Details for glycine-water solutions. . ......
5.2
System Details for glycine-water-methanol solutions ..........
.
47
.
47
......
55
.
56
5.3 Interaction Energy of water at the specific sites with the a-glycine
interface . . . . . . . . . . . . . . . . . . . . .
.... . . . . . . . . .
69
Chapter 1
Introduction
Understanding the control of polymorphism in organic crystals is of paramount importance to pharmaceutical and food industries and others, and is also of theoretical
relevance in solid state chemistry and physics. Significant variability in product performance in the chemical and food industry has been attributed to polymorphism,
and it continues to pose a challenge to pharmaceutical industries to produce drugs
of consistent quality. Polymorphism occurs when a molecule packs in different ways,
giving rise to two or more crystal structures. Organic compounds owing to their
size and floppiness result in various plausible polymorphs, sometimes with close thermodynamic stability. These polymorphs differ significantly in physical and chemical
properties, making polymorphic control in manufacturing absolutely necessary . Differences in solubility between crystal forms of a pharmaceutical can lead to differences
in bioavailability in solid dosage, if bioavailability is dissolution limited.
Crystallization from solutions is the most commonly used and critical step in
pharmaceutical manufacturing for polymorph selection. Cases with solvent affecting
the final polymorphic outcome have been widely reported [21, but the underlying
mechanisms still remain unclear. Research on polymorphism is fraught with unique
difficulties due to subtlety in polymorphic transformations. Exploring the molecular
self-assembly process during the nucleation process has proved very challenging, both
experimentally and computationally.
The growth synthon hypothesis [2] tries to establish a link between liquid phase
monlecular assemblies and their solid state counterparts in the final crystal, but excelptions to this rule exist. Tailor made additives [5] have been extensively used to
manipulate the final crystal outcome by inhibiting growth of certain polymorphs, but
the:,ir applicability has been found only in limited systems like glycine, alanine etc.
Thbe mechanisms through which solvent affects polymorphic selection still remains
uncit(rtain and much exciting work is to be done. The solvent selection process is
nmostly restricted to heuristic experimental screens, and any rational guidelines are
extlremely desired.
1.1
Objectives & Overview
The purpose of this thesis is to gain molecular level perspective into the solvent effect
on crystallization through the use of simulation techniques like Molecular Dynamics and Umbrella Sampling [6]. Glycine crystallization in water and water-methanol
mixtures is the model system chosen to study these effects. It has been observed that
glyci ne crystallizes in its least stable form when precipitated from water-methanol
solutions [4]. This system is investigated to better understand the underlying mechaniismns which could lead to rational selection of solvents.
The thesis is organized in six chapters. Chapter 2 discusses the nucleation and
growth processes in crystallization from solutions and the relevant literature review
of the solvent effect on polymorph selection during these stages.
In Chapter 3, the polymorphism exhibited by glycine and the effect of water and
methanol on the final crystal polymorph is discussed in detail. The hypothesis underlyinig the crystallization of the least stable /-glycine from water-methanol mixtures,
as proposed by Weissbuch et al.
141,
is discussed in this chapter.
It is our objective to test this hypothesis and gain molecular level insights into the
solv,ent effect on polymorph selection. For this purpose, molecular level simulations
of glycine in solutions and solvent-crystal interfaces were carried out.
Chapter 4
discusses the validation of force fields and parameters for glycine polymorphs. In
Chlapter 5 the results obtained from the solution and interface simulations of glycine
are presented and discussed in the light of the proposed hypothesis 14]. Chapter 6
summarizes the findings with major conclusions and suggestions for future work.
Chapter 2
Organic Crystal Polymorphism
2.1
Introduction
Polymorphism is the ability of a molecule to 'stack' into different molecular coinformations while retaining the same chemical composition [7]. In the pharmaceutical
industry, a very large number of pharmaceuticals exhibit the phenomenon of polyrrm,:rphism. 70% of barbiturates, 60% of sulfonamides and 23% of steroids exist in different
polymorphic forms [8]. The existence of polymorphism in the case of antiviral drug
Ritonavir has had dramatic commercial effect on pharmaceuticals. The manufacture
of Norvir (commercial name for Ritonavir) semi-solid capsules formulation involved
the preparation of a hydroalcoholic solution of Ritonavir which, although not saturated with respect to form I was 400% supersaturated with respect to form II. The
sudden appearance and dominance of this dramatically less soluble crystal form made
the formulation not manufacturable [9]. It was necessary to immediately reformul;ate
Norvir. These factors combined to limit inventory and seriously threatened the supply of this life saving treatment for AIDS. Another classic example is polymorphism
exhibited by compound ROY [1]. Six solvent free polymorphs of ROY are shown in
Figure 2-1. Glycine, a simple amino acid, can pack itself in three different crystal
structures.
Polymorphism and polymorphic transformations of organic systems have been
studied extensively since development of X-ray diffraction techniques. Earlier studies
L
r)
·4~t C
I
"Ch
~toy*V
4
HCH3
Figure 2-1: Polymorphs of ROY. [1]
mainly focused on characterization of crystal structures of various organic compounds
and methods of manufacturing these polymorphs [10]. Recent studies typically examnine conditions under which polymorphic transformations occur including humidity, pressure, solvents, additives and other process induced transformations [5, 11-14].
Ea:rlly computational investigations typically focused on ab initio polymorph prediction:i of various organic systems [15]. Recent studies on atomistic simulations of solutions and organic crystal interfaces provide molecular level insights into polymorphic
selection [16, 17].
Effect of solvents on organic crystal polymorphism has gained importance because the use of solution phases as media for homogenization and crystallization for
the subsequent assembly processes are common [2]. Understanding the mechanistic
role of solvents in polymorph selection is of great significance.
Solvent is an im-
portant consideration in solution crystallization, which affects the morphology, size
distribution, downstream processing, as well as polymorph of the final product. In
pharmaceutical industries, solvent screening is the first and the most important step
in polymorph study. Despite this, surprisingly little is known about the molecular
self assembly processes that surround the nucleation event and in particular the link
between solution speciation, molecular aggregation and the nature of intermolecular
interactions in the resulting crystal. This gap in understanding is particularly evident
in systems which exhibit crystal polymorphism, when small changes in solvent choice
and crystallization conditions can yield a new crystal structure. Several studies, both
experimental and computational, have been conducted to gain insights into the effect
of solvents on crystallization [16-18]. The process of crystallization from solution can
be divided into two stages: nucleation event to form an embryo, and the growth of
the embryo into a crystal. Solvent plays a major role during both these stages and
hence influencing the final crystal structure. In the following sections the relevant
literature on the solvent effect during nucleation and growth stages is reviewed.
2.2
2.2.1
Nucleation Controlled Polymorph Selection
Nucleation Theory
The process of nucleation has been the subject of almost continuous study over the
past 150 years [19]. The first notable observations were made by Ostwald concerning
the effect of sample volume, seeding and metastable states. Ostwald Rule of stages
[201 summarizes the complex interplay between thermodynamic and kinetic factors
in nucleation. Ostwald indicates that in a polymorphic system the crystallization
process starts with the appearance of the least stable of the known forms. This
is the first indication that the polymorphic outcome during crystallization is not
a fixed parameter rather it will be determined by the underlying kinetic processes
reflected by the crystallization pathway. Paracetamol, for example, crystallized from
acetone solution will always yield crystals of the stable monoclinic form I whereas
crystallization from the melt will always produce the orthorhombic metastable layer
structure, form II [21].
Classical nucleation theory originated in the 1930s and is largely attributed to
work by Becker, D6ring and Volmer [22,23].
The kinetic formulism was further
developed by Volmer [241. The basic idea is that thermal fluctuations give rise to the
appearance of small nuclei of a second phase and occasionally produce a long chain of
favorable energetic fluctuations, thereby creating a nucleus exceeding the critical size
[25]. Although this second phase has favorable lower free energy, there is a free energy
pen:)alty associated with the creation of an interface. Hence nucleation is an activated
pr:,cess in which the transition state is associated with an assembly of molecules held
together by intermolecular interactions and interface structure. The free energy AG,
of s(econd phase is the sum of a negative volume term and a positive surface term.
Forr a spherical nucleus, AG is given by
AG = - 7R3AG, + 47rR 2 7
3
(2.1)
where, R = radius of the nucleus, AG, = bulk free energy difference per unit volume
betl:ween the first and second phases, and y = surface free energy of the second phase
pe:-: unit area. The maximum of the free energy in 2.1 corresponds to the critical size
of the nucleus Rcritical, given by
Rcritical =
2AG,
(2.2)
Combining equations 2.1 and 2.2 produces an expression for the maximum free
energy barrier of nucleation, AGcritical,
4
AGcritical -=7ftR 2 critical
3
(2.3)
From 2.3, the rate of nucleation J, can be expressed using transition state theory,
J = A xp
AGcritical
(2.4)
wlhere, A = frequency factor, k = Boltzmann Constant, T = temperature.
IHowever, several fundamental limitations exist in classical nucleation theory. The
first is that the critical nucleus is treated as a bulk phase, and the surface is modeled
as atn infinite plane. These assumptions seemingly do not hold if the critical nucleus
is -:' the order of a few nanometers. In addition, classical nucleation theory does not
pr:,Nvide information regarding the formation of the critical nucleus. A nonclassical
th• !,ory of nucleation has been developed by Oxtoby et al. to overcome the limitations
posed in classical nucleation theory [26, 27]. Instead of treating the critical nucleus as
bulk material, the free energy of the second phase is quantified via density functional
theory. These types of non-classical nucleation theories serve to overcome, albeit
qualitatively, a number of shortcomings of classical nucleation theory. However, these
models still do not accurately capture the process of nucleation [27].
The kinetic theories were developed to explain macroscopic phenomena at tirie
when the available experimental techniques were restricted to a resolution of 1.0 nmln
at best. The advent of modern techniques and methodologies, including atomic probe
microscopes, neutron diffraction and scattering and molecular modeling, have led us
routinely to expect resolution and visualization down to 0.1 nm, an advance which
has only just begun to be reflected in experimental and theoretical studies [28, 29] of
the nucleation process. There is growing evidence that nucleation from solution is
itself a two-step process. The first step involves the formation of a liquid like cluster o(
solute molecules, while the second involves the reorganization of such a cluster into an
ordered crystalline structure [30-32]. ten Wolde and Frenkel considered solute-sollute
interactions given by a modified Lennard-Jones potential and proposed the presence
of a metastable fluid-fluid critical point close to which the free energy barrier for
crystal nucleation is strongly reduced and the nucleation rate increases by many drders of magnitude [301. Anwar and Boateng showed by molecular simulations that
crystallization in highly supersaturated systems involves liquid-liquid phase separation followed by nucleation of the solute phase [31]. By using molecular dynamics to
study the nucleation of AgBr in water, Shore et al. have provided further evidence to
the conjecture that nucleation of crystals from solution may be a two-stage process
[321. They have further suggested that disordered clusters are actually the stable
state until the cluster size becomes rather large. Using small-angle X-ray scattering.
Chattopadhyay et al. have directly studied the nucleation of the amino acid glycine
from its aqueous supersaturated solution [33]. Their results are consistent with a
two-step nucleation process.
Even with the advent of these techniques, determination of the structure of the
critical nucleus has proven to be a daunting task in simple single component sys-
temns like water [34], nitrogen [351 and carbondioxide [36]. This problem still remains
unsolved for systems involving two or more components, for e.g. a solution. Even
though the structure of critical nucleus remains unresolved, an important result deduced through the use of these advanced techniques is the Growth Synthon hypothesis 121 that connects the solvent-solute interactions to the crystal structures obtained.
This is explained in the following section.
2.2.2
Growth Synthon Hypothesis
Crystal engineering has spawned the notion of the structural synthon and growth syntho'ns. The structural synthon refers to the hydrogen bond structure of the building
blocks in the crystal packing motifs and have been derived from the vast amount of
crystal structure data available. The growth synthons (or 'growth units') refer to the
cotnformations exhibited by the solute molecules in the solutions. The understanding
of :;he link between the growth unit and the structural synthon then becomes the basic structural problem for nucleation theory. The growth synthon hypothesis suggests
that the most stable growth synthon has the highest probability to pack into crystal
as the structural synthon which can be seen using XRD.
M
M
Figure 2-2: The link between self-assembly in solution and its crystal structure.
The growth synthon hypothesis was successfully used to explain the polymorphic
behavior of tetrolic acid. Davey et al. [37j detected the presence of carboxylic dimers
formed by two tetrolic acid molecules in chloroform using FTIR, which is consistent
with the dimer structure (structural synthon) present in the crystal obtained fr,.,m
chloroform solution. They also found that, when ethanol used as the solvent, no
dimer structure can be seen in solution, which is consistent with the fact that no
dimer structure is present in the crystal obtained from ethanol. The relation betweeh
the growth and structural synthon for tetrolic acid is shown in Figure 2-2.
Ar
Aw
d
I
ii
N0416"T
Fr
"
I
i..
·
I
a
;~
~·
*
;·=e
;r
;,~
~
f
'
;g
(b)
Figure 2-3: Two polymorphic forms of 2,6-dihydroxybenzoic acid: (a) the metastable
form 1 dimer structure, and (b) the stable form 2 catemeric structure [2].
A Similar trend was observed in the case of 2,6-dihydroxybenzoic acid. The a::id
exists in only two known polymorphic forms: form I is a metastable monoclinic for tri
based on centrosymmetric carboxyl dimers [38] while the stable form II is noncentxvric
based on a catemer motif which utilizes a carbonyl-hydroxy interaction [39]. Figure 23 shows the structural elements of these two structures. In this case it was found that
in quiescent solutions the metastable dimer structure nucleated preferentially from
toliuene, whereas in chloroform direct nucleation to the stable form was possible.
A combination of solubility data, UV/vis spectroscopy and calculation of solvation
enildthalpies all pointed to the preferential assembly of dimers in toluene solutions and
catcemers in chloroform, yielding again a clear link between the assembly process in
s(!lution and the nucleation event 140]. A further example is the case of glycine. From
a(itieous solutions, at the isoelectric pH of 6.1, glycine crystallizes in its metastable
cenit:rosymmetric a form. The consistent appearance of a-glycine has been explained
by the existence of centrosymmetric zwitterionic dimers in aqueous solution which
naturally pack to yield the a structure [41].
However, limitations to the growth synthon hypothesis still exist. For example
Im.nldelic acid crystallizes in a dimer based form from its racemic solution, although
no, dimer based growth synthon can be detected in solution [42]. Even there exists no
direct data that relate to the structure of the molecular clusters that are smaller than
crit-:ical size. The growth synthon hypothesis provides some evidence that molecular
as,:ý'emnbly in the liquid phase can mirror the packing of the potential polymorphs in a
syternm. The nature of the solvent-solute interactions, however, can play a significant
ro!le in determining the viability of these clusters and hence direct the structural
outIcome of the crystallization. The understanding of the link between the growth
n:it and the structural synthon remains the basic problem for nucleation theory and
r:I:I(h exciting work remains to be done.
2.3
2.,3.1
Growth Controlled Polymorph Selection
Crystal Growth in Solutions
It was very well known that solvent affects the crystal shapes, i.e., the ratio of growth
rates of their various faces [17]. The effects of solvent on crystal growth have been
ex-lý.ensively studied for over a century. Historically it was probably the work of Wells
[431 which first drew the attention to the importance of solvent in determining growth
morphology. Using resorcinol as an example he illustrated the significance of a solvent
specific interaction with different crystal faces. According to Wells, the faces ternminating in hydroxyl groups of resorcinol are blocked by water molecules. This leads
to preferential development of these faces during the formation of resorcinol crystarls
The nonpolar solvent benzene does not affect this selective adsorption. The crystal
grows symmetrically in opposite directions. Other studies followed Well's work and
in particular the works of Watson [44] and Kleber and Raidt [45] are worth noting.
These works were the first to link growth behavior to solution chemistry.
In general, the crystal growth kinetics from solutions is governed by two factors
relating to the nature of the growing interface [18]. One is related to the degree of
molecular roughness as determined by the energetics of step creation at the surface
and the other is concerned with the stereo and thermo chemical nature of adsorption
of solvent on the surface [46].
Surface Roughness: In the statistical mechanics description of the interface between a solid and fluid phase the roughness is quantified in terms of the a
factor [46]. The a factor defines the enthalpy change which takes place when
a flat interface is roughened [47].
The value of a can be related to the ex,
pected growth mechanism of a face. According to the surface roughness theory
if a < 3, the interface is rough, and the growth is linearly dependent on the
supersaturation. If a is greater than 4, the interface is smooth, and the grow.th
occurs at the steps generated by defects of surface nucleation.
Surface Adsorption: In terms of the crystal growth process it is important to
consider the nature of the interaction between individual solvent and sollute
molecules since incorporation of a solute molecule in a growth site requires desorption of solvent molecules. For certain combinations of solvent and solute
interfaces it is conceivable that there could be strong adsorption of solvent in
the growth sites hence reducing the growth rate of that interface. No general formalism exists to describe the relationship between adsorption and growth rate.
Each example has to be considered in turn due to the specificity of molecular
packing in the solid phase [181. When the solvent-solute intermolecular interactions are strong the solute molecules are solvated and the growing surfaces of
the crystals are covered by solvation layers which must be removed prior to the
deposition of additional layers. The degree of solvation which is determined by
the structure of the surface layer may vary from one face to another.
A(HA(HO)
A(H 0)r
I
H2O)' k
A(I
3'
Figure 2-4: An illustration of various processes involved in the crystal growth process
frp:mr aqueous solutions [3].
The above two factors are qualitative indications of the growth mechanisms and
solvent effects on growth interfaces. A more general formulism for crystal growth from
solutions is developed by Bennema [3] by adapting the Burton, Cabrera and Frank
surface diffusion model [48]. According to this theory, the first step in crystallization
from solutions is desolvation of the crystallizing solute molecules and the surface sites,
avl:i then the entrance of the solute molecule into the solvation layer of the faces.
Thbis is followed by surface diffusion until a step is reached permitting incorporation
int:o the crystal lattice at a kink site. The activation free energies associated with
these respective steps are AGdeh (dehydration activation free energy for entering the
surface layer), AGsdiff (activation free energy for making a diffusion jump from one
eqiuilibrium position to a neighboring one in the surface layer) and AGkink (activation
free dehydration energy for the entry of growth units into kinks from the surface layer).
Figure 2-4 illustrates the various processes involved in the crystal growth process from
aqueous solutions.
It is qualitatively important to understand the free energy barriers that dominate
the growth process.
AGdeh and AGsdiff are both dependent on the interactions
between the solute and solvent molecules. It is generally true that AGsdiff is much less
than AGdeh for crystal growth from solutions. All three free energies are different on
different faces. For growth from the vapor phase AGdh = 0 and the rate determining
step is AGkink. It is also important to note that for layer growth mechanism (where
the creation of kink is not necessary), AGdeh determines the rate of crystal growtfh.
Therefore, for growth from the solution, strong interactions between solute and solvent
at specific crystal faces may lead to AGdeh >> AGkink, and thus the desolvation of
specific crystal faces becomes the rate determining step.
2.3.2
Crystal Growth in Polar Organic Crystals ("Relay Type"
Growth Mechanism)
A very interesting case with an entirely different approach towards crystal growth
from solutions is observed in the growth of polar organic crystals like y-glycine aild
(R-S)alanine. These two crystals have similar packing features and only the growth of
7-glycine is discussed here. -y-glycine in solutions has a flat (001) face perpendicular
to the polar c-axis at one end and capped faces at the opposite end as shown in
Figure 2-5 [49]. According to crystal growth and etching experiments [50, 511 the
C02- groups are exposed at the (001) face, the "flat -c end", while the NH3 + amino
groups are exposed at the -c capped end [50,51]. Various experiments involving
the comparison of the relative rates of growth and dissolution of the crystals of (1IS)alanine and -y-form of glycine indicate that in aqueous solutions the -c carboxylate
end of the crystals grow and dissolve faster than the +c amino end.
Inspection of the packing arrangement of -- glycine Figure 2-6 reveals that (001)
carboxylate faces comprise regular pockets on a molecular level and can be regarded
as corrugated in two dimensions. The binding of water in these pockets can be
qualitatively explained by looking at the water-glycine interactions in the pocket.
The water molecules inside the pocket can essentially take two different orientations;
one orientation comprises of O-H. --O hydrogen bond and two 0-0 lone-pair-lone-
(001)
4:~~44~
~
(013)
Figure 2-5: Packing arrangement of y-glycine delineated by crystal faces, as viewed
down the b-axis. The capped faces at the +c end of the polar axis expose NH +
3
groups at their surface the opposite faces expose C02- groups.
pa~ir repulsions and the other two O--H... O bonds and one 0-0 lone-pair-lone-pair
repulsion. Consequently introduction of water yields repulsive or at best weakly
attractive interactions. The pocket will therefore be unhydrated or slightly hydrated
arI:1 relatively easily accessible to approaching solute molecules as shown in Figure 26. In contrast, the water molecule may be strongly bound to the outermost layer of
C()2- groups via O-H-.. O (carboxylate) hydrogen bonds.' As glycine molecules
are incorporated into adjoining pockets, the CO2- groups of newly added substrate
molecules will expel the water bound on the outermost surface, thereby generating
newn unsolvated pocket on the crystal surface. This relay process of solvent water
binrding and expulsion helps growth and dissolution by both desolvating the surface
ari:d perpetuating the natural corrugation of the surface, at a molecular level (Figure 2A general relay type mechanism is depicted in Figure 2-7. The difference between
twio types of sites (A & B) is emphasized by assuming a corrugated surface such that
thc A-type site is a cavity and the B-type site is on the outside upper surface of the
cavity. Figure 2-7(a) shows the B-type sites blocked by solvent S and the A-type
sites unsolvated. Thus solute molecules can easily fit into A-type sites. But once
docked into position (Figure 2-7(b)) the roles of the A and B-type sites are essentially
' The water molecules bound to the outermost layer of groups via O-H-O hydrogen bonds
provide
additional stabilization energy to the (001)surface layer which has a low molecular density
with
all
mol ecular dipoles pointing approximately in the same direction.
rpp~apein~e
b•
w
(00))
Figure 2-6: Packing arrangement of y-glycine as viewed down the b-axis, depicthing
the relay mechanism at (001) interface. The bound water molecules are ejected from
the pockets as glycine from solution approach the interface.
reversed and the solvent molecules which originally were bound to B-type sites wonild
be repelled since they now occupy A-type sites. This cyclic process can lead to f;ast
growth. In such a situation described here, where desolvation is rate limiting, it. is
implicitly indicated that the free energy of incorporation of a solute molecule helps
to displace bound solvent.
approaching
solute to site A
solvent
Ssolutebound
0
00
(a)
-
rejected
,solvent
nw
fcie
000
(b~)
Figure 2-7: A schematic representation of "relay type" growth mechanism.
These models help in understanding the polymorph selection and morpho.olo.gy
control of the final crystal. The Growth Synthon hypothesis helps in analyzing the
polymorphic selection in some systems. The crystal growth models are also helpfuill
in i•:xplaining the inhibition of certain crystal forms even though they are the most
stable under the given experimental conditions. Hence it is important to realize that
polymrorph selection in solvents can occur at both nucleation and growth stages. The
nex,:t chapter introduces the solvent effects on glycine crystallization. This thesis is
pri marily concerned with using the above mentioned theories in understanding glycine
poly.morph selection in water-methanol solutions.
Chapter 3
Glycine Polymorphism
3.1
Glycine and its Polymorphs
The glycine model system is explored in this study to gain molecular level insighl:ts
into the solvent effects on polymorphism. Glycine is not only chosen for its molecular
simplicity and the abundance of experimental results in the literature but also due to
its interesting polymorphic behavior in solutions. Three crystalline polymorphs were
described for glycine: two monoclinic (a, s.gr. P21/n, and
f,
s.gr. P21) and one
trigonal (y, s.gr. P31). The three polymorphs differ with respect to the connectivity
between zwitterions (NH3+-CH2-COO-) [52].
The first attempts to study a-glycine by means of X-ray diffraction were undertaken by Bernal [53] and by Hengstenberg and Lenel [54] independently. The
explicit crystalline structure was refined by Albrecht and Corey [55], and a precise
refinement of not only the heavier atoms but also locating of the hydrogen atoms for
a-modification was carried out by Marsh [561. In the a-polymorph the zwitterions
are linked by hydrogen bonds in double anti-parallel layers, the interactions between
these double layers being purely van der Waals as shown in Figure 3-1.
/-glycine had already been obtained and described by Fischer [57] at the beginning,
of the century, whereas X-ray examination was performed quite a long time later [58].
It was suspected that this fact was due to the general low stability of this phase
and possibly due to irreversible transformation into a or -y-glycine forms. In the
+--4s--+--4
C
Figure 3-1: a-glycine as viewed down the a-axis. Crystal consists of hydrogen bonded
bilayers.
,3-ptolymorph individual parallel polar layers are linked by hydrogen bonds in a threedimensional network (Figure 3-2).
b
C
Figure 3-2: 3-glycine crystal structure as viewed along the a-axis illustrating the
three-dimensional hydrogen bond network.
The structure of y,-glycine was first revealed and resolved by X-ray diffraction
by litaka [59-611. Moreover, he analyzed not only the packing of the molecules in
the crystal lattice, but also the role of the hydrogen bonds in the formation of the
fra-mework and the substructure of the hydrogen bonds. Investigations of 7-glycine
at 298 and 83 K were carried out by means of neutron diffraction by Kvick [62] in
order to determine electron density changes of the molecules with temperature. The
-y-polymorph consists of polar helices linked with each other in a three-dimensional
polar network (Figure 3-3).
b
a
.
(a)
Figure 3-3: 7-glycine crystal structure.(a) viewed down c-axis, (b) viewed parallel to
c-axis. The crystal consists of polar helical hydrogen bond structure
The y-polymorph is the most stable form at ambient conditions, although the
a-form crystallizes much more readily, and the a-form (with rare exceptions) was n:ot
observed to transform into the y-form at these conditions. With increasing temperature, the order of stability inverts, the a-form becomes the most stable one abo(:e
-440 K, and a y --+ a polymorph transition is observed when the 7-form is heat,: d:.
On subsequent cooling, the a-form does not transform back to the -y-form, presuniably due to kinetic reasons. The 0-form is obviously metastable at all temperatures
[53].
3.2
Solvent Effects on Glycine Polymorphism
Experimentally, the unusual feature of this system is that crystallization from aqueous
solution at the natural isoelectric pH (5.97) always gives the a polymorph and it seems
that under these conditions 7 never appears, despite its thermodynamic stability. It
has been previously concluded that the nucleation and apparent 'stability' of the
metastable a form at pHs close to the isoelectric point is a reflection of the presence
of centrosymmetric dimer 'growth units' (see section 2.2.2) in solutions. On the basis
of solution [63], interfacial and solid-state chemistry [64] it was suggested that at
alld around the isoelectric point, glycine is dimerized in solution as centrosymmetric
paý:irs of zwitterions. Myerson and Lo predicted that glycine exists mostly as dimers
in :~supersaturated solutions by measuring the diffusion coefficient for supersaturated
aquecous solutions of glycine [65]. Using small-angle X-ray scattering, Chattopadhyay
et :l. have directly studied the nucleation of the amino acid glycine from its aqueous
sui, ersaturated solution and indicated that glycine molecules exist as dimers in the
sup:)ersaturated solution [33]. It is then apparent that nucleation from such solutions
(co:uld lead directly and spontaneously to the metastable a structure [14].
This is
furi:ther supported by the experiments involving S-control (supersaturation control) by
Ch:ew et al. where they have shown that a-glycine grows about 500 times faster than
"7-lE:
ycine in neutral aqueous solutions. This difference in growth rate corresponds
to - difference in activation energy for growth of -15kJ mol-' calculated from the
Ar•ihenius equation.
This large difference in activation energy is attributed with
th(::, dissociation of Glycine dimers in solution prior to growth of 7-glycine, but their
prc•servation in the a-glycine crystal structure [66].
TFihe next interesting issue is why 7-glycine forms at low pH aqueous solutions.
In, each of its polymorphic forms glycine molecules pack as zwitterions. As discussed
abl:ove, the 'stability' of the metastable a form at pH close to the isoelectric point is a
reflection of the presence of centrosymmetric dimner 'growth units' in solutions. The
efli!t of moving the pH away from the isoelectric point is in reducing the proportion
of t:he a-form 'growth unit' (because singly charged glycine molecules will not form
cy,:tic dimers). This would then increase the proportion of monomeric zwitterions
ava,:,ilable to form the polar chain structure of the y-polymorph.
In the following thesis, precipitation of the least stable 3-glycine from (50% v/v)
me:tlý.hanol-aqueous solutions has been addressed from a molecular standpoint. The
exlj;erimental results and underlying hypothesis (from literature) is discussed in the
n:::t section.
3.3
Glycine Polymorph Selections in Water-Methanol
Solutions
As discussed previously, a-glycine crystallizes primarily in aqueous solutions through
hydrogen-bonded cyclic dimer growth units. The precipitation of y-glycine at low
or high pH aqueous solutions has been explained successfully by Davey et al. 167].
Another interesting case of polymorphic selection of glycine is the precipitation of 3glycine in alcohol-water solutions. The conundrum that the more thermodynamically
stable a- and y-glycine polymorphs do not generally precipitate in aqueous solutions
containing methanol or ethanol under the specified experimental conditions was addressed from the growth kinetics of the three polymorphs of glycine coupled with an
analysis of the action of the solvent at the various crystal faces by Weissbuch et a1l.
[4].
3.3.1
Precipitation of 3-glycine
The first crystallization of 0-glycine from water-alcohol solutions was reported by
Fischer [57]. The crystal structure [58] is polar (space group P21) and comprises
hydrogen-bonded layers, which are similar to those observed in the a form, but
which are interlinked by NH... 0O and CH- - -O interactions through a twofold screw-ý
symmetry axis perpendicular to the layer plane (Figure 3-2). The addition of alcohol
reduces the solubility of glycine from 25.0 g/100 mL water (25 'C) to 2.65 g/100 niL
solvent in 50.1%(v/v) ethanol-water mixtures. It was hypothesized that this reduced
solubility would result in an increased concentration of solvated glycine monomers
relative to that of hydrogen-bonded cyclic dimers. Such behavior is apparently coaisistent with the preferred precipitation of 3-glycine from alcohol-water solutions because the crystal structure consists of hydrogen-bonded monomer units, as opposed
to a-glycine which comprises cyclic hydrogen-bonded pairs.
Long needles of 3-glycine were grown in water-ethanol mixtures containing 50,
26.1, and 10% (v/v) ethanol and also from 1:1 water-methanol mixtures containing
FiVgure 3-4: Packing arrangement of @-glycine. The (010) "azure" and (010) "pink"
surfaces, are exposed at the interface.[4]
4.0, 19.0, 35.9, and 5.0 g glycine/100 mL solvent, respectively. Growth kinetic measurements of single 0-glycine crystals in 1:1 water-ethanol solutions at 25 'C reveal
a ftst growth at one pole of the needle and a very slow growth at the opposite end.
The absolute polarity [68] of 0-glycine was determined by employing "tailor-made"
additives[5], in this case racemic tryptophane (Trp). It was concluded that 0-glycine
gro:,ws faster at the side with exposed C-H bonds (colored azure) than at the opposite side with exposed N-H bonds (colored pink; Figure 3-4). Previous studies have
sh(iwn that the relative rates of growth at the opposite ends of polar crystals in polar
solvents can be correlated directly with the relative rates by which solvent molecules
are stripped from the opposite ends [49, 69-71]. The faster growth rate at the 0glycine pole with exposed C-H bonds is in agreement with this model; the water or
ale:,hol solvent molecules can be attached more effectively to the slow growing glycine
surface with exposed N-H bonds through strong OH,, ... 0y-
and NHgly+... Oso
interactions than to the fast growing 0-glycine pole with exposed C-H bonds with
strong OHor ... O9ty- interactions but only weak CHgl ... 0,,o interactions.
3.3.2
Inhibition of 'y-glycine
The absence of the stable 7-glycine form in crystals formed in alcohol-water solutions
is explained by examination of its growth properties (see section 2.3.2). The polar
~th~arusf
Figure 3-5: Packing arrangement of 7-glycine showing the pockets of fast growing
(001) face that are poisoned by the adsorption of ethanol and methanol molecules
(shown as "balls and sticks").[4]
crystal structure of y-glycine (space group P31; Figure 3-3), which is not composed
of cyclic glycine pairs, is delineated by a (001) face at which C0 2 - groups emerge and
capped crystal faces at the opposite end that expose NH3+ groups. Previous studies
[72] have shown that -y-glycine, grown in aqueous solutions and in the presence of
auxiliaries that inhibit the crystallization of a-glycine, appear as [001] needles that
grow along the polar c axis much faster at the end of the crystal with the CO)2-
groups than at the opposite capped end. This unidirectional growth was interpreted
in terms of a "relay" mechanism as described in section 2.3.2. However, ethanol and
methanol solvent molecules can reside within the pockets through OH,,1 ... Ogy-
and CHalcohol ... Ogly- hydrogen-bonding interactions, thus inhibiting growth at the
C(2- end of the crystal (Figure 3-5). In several of the crystallization experiments
carried out in water-ethanol mixtures, the few -y-glycine crystals that were observed
exhibited morphology in keeping with the proposed inhibition by ethanol or methanol
of growth along the otherwise fast growing C02- end of the crystal (Figure 3-6).
I
-. :::
9
Fig;ure 3-6: y-glycine crystals observed in several crystallizations from a 1:1 waterethanol solution.[4]
3.3.3
Inhibition of a-glycine
The surface of the fast and slow growing ends of a-glycine are very similar in structure
to the {010} surfaces of 3-glycine with either exposed C-H or N-H bonds, as
shown in (Figure 3-7) respectively. On the basis of the realistic assumption, which
is supported by experimental evidence [64,73], that glycine molecules in aqueous
solution dock onto the crystal surface primarily as hydrogen-bonded cyclic glycine
pairs, it is thought that a {010} face will expose the faster growing surface with
exposed C-H bonds to a much larger extent than the slower growing surface with
exposed N-H bonds.
4LA
(a)
(b)
Figure 3-7: Packing arrangements of ca-glycine (a) exposing weak solvent binding C H bonds to the solution at (010) surface (azure) or (b) exposing strong solvent-binding
N-H bonds to the solution at the (010) surface (pink).[4]
Figure 3-8: a-glycine crystals grown from 9:1 water-ethanol.[41
It was anticipated that reduced solubility of glycine in solution caused by :he
presence of alcohol would lead to a higher proportion of solvated glycine monomer
units docking onto the a-glycine {010} surface sites with exposed N-H bonds. Thus,
the time required to strip the overlying solvent molecules, prior to formation of the
glycine cyclic dimer growth units and propagation of the glycine bilayer with exposed
C-H bonds on its {010} surface, would lead to an overall reduction in growth rate
along the ±b directions of the a-glycine crystal. Indeed, the a-glycine crystals obtained from a 9:1 water-ethanol solution tended to display more well developed {0()1i0}
faces (Figure 3-8) than crystals obtained from purely aqueous solutions. Therefore,
enm:bryonic crystallites would expose slow-growing {010} surfaces at higher concentrati on's of the alcohol in contrast to /3nuclei, which has only one slow-growing polar
end:/ and so results in a preferred kinetic precipitation of the latter. Thus, water or
alc:l:Iol as solvent impedes growth normal on the {h01} faces of the needle crystals
as a result of strong solvent attachment to these faces through OH8 o,". 0,,y- and
N:l•' +
---Ool
hydrogen bonding interactions.
Chapter 4
Validation of Force Field Parameters
It is necessary to have the right force field and parameters for performing molecu-lar dynamic simulations of glycine solutions and interfaces. The force field consists
of intramolecular parameters (defining the bonds, angles, dihedrals and impropers),
Lennard-Jones terms for the dispersion energy between atoms and the partial charges
for electrostatic interactions. The validity of a particular force field and the pararmreters has to be tested both for solid state and solution behaviour. For solid state, the
lattice parameters of the three glycine polymorphs and the enthalpy of sublimation are
compared with their corresponding experimental values. For solution behavior, water
coordination numbers and distances are calculated and compared with experimental
values obtained from neutron diffraction measurements.
The experimental crystal structures and lattice parameters of the three glyc:iie
polymorphs are obtained from the Cambridge Structural Database (CSD) [74J. The
experimental lattice parameters of glycine polymorphs are tabulated in Table 4.2.
Using the coordinates obtained from the CSD, minimization and dynamics of the
polymorphs were performed in NAMD [75] with periodic boundary conditions using
CHARMM [76], OPLS [77], and AMBER [78] force fields. Ewald summation [791. is
employed for the electrostatics. The number of unit cells and atoms that comprise the
supercell used in periodic boundary conditions is shown in Table 4.3. Only OPLS was
successful in maintaining a stable crystal structure of all the three polymorphs. AMBER and CHARMM failed to maintain the stability of a and y glycine polymorphs.
The OPLS force field was further tested for its ability to reproduce experimental lattice parameters at room temperature using molecular dynamics simulations. Constant
Pressure Molecular dynamics simulations of all the three polymorphs were carried out
for Ins at 300K and at 1 atmosphere pressure without applying any constraints on the
sujpercell. Even though OPLS force field was successful in reproducing stable hydrogen bond networks in glycine polymorphs, the lattice parameters of a-glycine showed
large deviation (- 8%) from experimenatally observed parameters. This is resolved
by modifying the lennard jones parameters of the a-hydrogen (H4 and H5) attached
to a carbon (Cl) in glycine zwitterion (Figure 4-1). The lennard-jones parameters
for a-hydrogen developed by Veenstra et. al. [80] and implemented in AMBER forcefileld for glycine zwitterion were used in this study. The OPLS LJ parameters and
the!, parameters used in this study are tabulated in Table 4.1. Shown in Table 4.4 are
th(e:: root mean square deviations (RMSD) of the minimized crystal structures using
the modified OPLS parameters. Table 4.5 shows the percentage deviation in lattice
parameters of the three polymorphs during the 1 nanosecond simulation. The crystal structures and the hydrogen bond networks in the polymorphs were successfully
maintained during the dynamics.
Figure 4-1: Glycine zwitterion with all the atoms labeled. The lennard-jones paramneters of H4 & H5 were modified to obtain a better agreement with experimental
observations.
KParameters
Table 4.1: Lennard Jones parameters for a-hydrogen.
,Ein (kcal/mol)
()PLS
PIModified [80]
-0.0300
-0.0157
Rmin A
1.4031
1.1000
Polymorph
Table 4.2: Polymorph Details for Glycine
Lattice Type
Lattice Parameters
7-glycine
Trigonal P31
a-glycine
Monoclinic P21/n
O-glycine
Trigonal P21
a = 7.046, b = 7.046, c = 5.491
a = 90, 8 = 90, - = 120.
a = 5.1054, b = 11.9688, c = 5.4645
a = 90, 0 = 111.70, 7 = 90.
a = 5.0932, b = 6.270, c = 5.3852
a = 90, 4 = 113.19, 7 = 90.
Table 4.3: Number of Unit Cells and molecules per super cell during minimizati( n
runs.
Polymorph
gycine
__-glycine
Number of Molecules
in the supercell
Supercell Size
_
5X5X6 unit cells
A
450 molecules
30.63X35.91X32.79) A3
7X5X7 unit cells
'35.65X31.36X37.70) A3
432 molecules
(35.23X35.23X32.95)
AXYR
uY
"n4t
cell
490 molecules
Table 4.4: Root Mean Square Deviation (RMSD) of glycine polymorphs during minimization using OPLS force field.
RMSD
Polymorph
a-glycine
0.393
0.285
,-glycine
y-glycine
0.236
Table 4.5: Percentage Deviation of lattice parameters of the three polymorphs of
glycine during a Inanosecond molecular dynamics simulation at 300 K with OPLS
force field
Polymorph
a-glycine
% Deviation of Lengths
a = -2.09%,
% Deviation of Angles
a = 0%,
3-glycine
b = 2.803%, c = -3.74%.
a = -2.9%,
b = 4.42%, c = -4.14%.
+0.35%, 7 = 0%.
a = 0%,
0 = +0.5%, 7 = 0%.
gcne
3=
a = -0.63%,
a = 0%,
b = -0.66%, c = 2.64%.
P = 0%, - = 0%.
The lattice energies of the glycine polymorphs at Zero Kelvin are tabulated (Table 4.6). The difference in potential energies obtained are qualitatively in good agreement with those calculated using DFT [811. The calculated enthalpy of sublimation
Table 4.6: Relative lattice energies of glycine polymorphs obtained by annealing the
crystals to Zero Kelvin.
rPolymorph
ar-glycine
.-glycine
ý--glycine
Lattice Energy(kcal/mol)
-176.32
-175.24
-175.14
Relative Energy (kcal/mol)
0.0
+1.08
+1.18
for glycine using the method illustrated by Bisker-Leib and Doherty [82] is ~ 34.3
kc.ial/mol. This is in close agreement with the experimental enthalpy of sublimation which is 32.6 ± 0.5 kcal/mol. The OPLS parameters with modified hydrogen
le:nnard-jones parameters were further tested for their ability to model glycine-water
intleractions. The radial distribution functions are calculated for selected atoms in
glycine with the oxygen atoms (O,) of water molecules. Kameda et al. [83] performed neutron diffraction measurements and showed that the ammonium group of
glycine in 5.0 M aqueous solution is coordinated to 3.0 + 0.6 water molecules and
tho! distance between nitrogen atom in glycine and oxygen atom in water is 2.85 ±
A. g(r)
0,5
between nitrogen and oxygen atom in water is shown in Figure 4-2. The
OII'LS parameters with modified a-hydrogen parameters give a coordination number
of
-
3.0 and a Ng,-O, distance of - 2.89 A. Also shown is the g(r) (Figure 4-3)
berween the ammonium hydrogen of glycine and the oxygen atom of water.
Similar calculations were performed for the carboxylate group of glycine. A coordin ation number of - 4.9 and a Ogly-O, distance of - 2.67 A is obtained (Figure
4-4). A previous ab initio study [84] found that the number of water molecules in the
firit hydration shell is - 4.7. A good agreement between the computational and experimental results bolsters our confidence in the interatomic force fields used. These
parameters are a good fit to model the solid state, interfacial and solution behavior
of glycine. Tabulated (Tables 4.7 and 4.8) below is a comparison between coordination numbers and distances calculated with the available experimental and ab initio
results.
Table 4.7: Average Coordination numbers with oxygen atoms in water.
Atoms
Calculated
Experimental
Ab Initio
0
NgEy- O
3.0
3.00 + 0.6 [83]
3.0 [84]
4.9
Ogly- O,
4.7 [841
Table 4.8: Average Atom - Atom Distance
Atoms
NgI-- Ow
Ogly- O
Calculated
2.89
2.67
Experimental
2.85 + 0.05 [83]
-
Ab Initio
2.75 [84]
RDF of Nitrogen(glycine) -- Oxyven(water)
Distnace (Angstrom)
Figure 4-2: Radial distribution function calculated for the nitrogen atom in glycieine
and oxygen atom in water.
LL
CI
Distnace (Angstrom)
Figure 4-3: Radial distribution function calculated for the hydrogen atom attached
to nitrogen in glycine and oxygen atom in water.
RDF of Oxygen(glycine) -- Oxygen(water)
LL
£3
Distnace (Angstrom)
Figure 4-4: Radial distribution function calculated for the oxygen atom in glycine
and oxygen atom in water.
Chapter 5
Simulation Methods & Results
5.1
Solution Behavior of Glycine
The hypothesis suggested for p-glycine crystallization from water-methanol solutions
by Weissbuch et al. revolves around the assumption that non-dimer like units deck
onto the (010) interface of a-glycine during its growth. This exposes the N-H groups
at the interface which further inhibit crystal growth of a-glycine. Simulations of
solution behavior of glycine in water and water-methanol solutions have been carried
out to explore the possibility of decrease in dimer like structures ('growth units' for
a-glycine). Initially, the stability of a dimer structure is tested by calculating the
lifetimes of a glycine dimer unit in water and water-methanol solutions. Free energy
of the dimerization reaction was then computed by the Umbrella Sampling method.
Further simulations were carried out to compute the fraction of glycine molecules
that exists as monomers and dimers in water and 50% v/v water-methanol solutions
with varying supersaturations.
5.1.1
Dimer Lifetime
Simulations with glycine dimer structure (shown in Figure 5-1) as the starting configuration were carried out in water and water-methanol solutions. The concentration of
methanol in the solution is 0.3088 mole fraction (which corresponds to 50% v/v water-
methanol mixture). Constant Volume and Temperature (NVT) Molecular Dynamics
sin:mulations in a 28 A box were carried out for 0.5 nanosecond. Four simulations
each in water and water-methanol solvents with different initial velocities were carried out to explore the stability of the dimer structure. The average of distances
dl and d2 (Figure 5-1) is plotted with time. The plots suggest higher lifetime of
the•! centrosymmetric dimer structure in water-methanol solutions. But to be able to
make any accurate conclusions about the relative stability of the dimer structure in
water-methanol mixtures it is necessary to calculate the relative free energies.
Figure 5-1: Dimer Structure of Glycine.
Figure 5-2: Illustration of lifetimes of Dimer Structure in Water. The dark line at
3.75 A represents the average (dl + d2)/2 for the dimer structure.
in water methanol mixtures
C14
_
_
-o
bfi
00
va)
Figure 5-3: Illustration of lifetimes of Dimer Structure in Water-Methanol mixturqes.
The dark line at 3.75 A represents the average (dl + d2)/2 for the dimer structur,.
5.1.2
Free Energy Calculations
Potential Mean Force (PMF) calculations were performed to compare the relative
stability of a dimer in water and water-methanol mixtures. MD Umbrella Sampliilg
[6] was used to obtain a free energy profile for the dimerization reaction. Shown below
in Figure 5-4 is the free energy diagram for the formation of a dimer structure with
two hydrogen bonds from glycine monomers. The reaction coordinate was assum,:ed
to be the average of the distances dl and d2 shown Figure 5-1. A harmonic umbrella
potential (k = (k o/2)(6 - 6,)2) is applied along the reaction coordinate at an interval
of 0.25 A. The system consisted of two glycine molecules in a 28 A cubic box. Constant
Volume and Temperature Molecular Dynamics were performed at 300 Kelvin.
As can be observed from the free energy plots the centrosymmetric dimer str-icture is more stable (- 1 kcal/mol) in water-methanol solutions. Even the activation
energy for dissociation of the dimer structure is higher (-0.2 kcal/mol) in 50%~cv
solution of water-methanol than in pure water. Even though the magnitude of the differences in free energies is relatively insignificant compared to the thermal fluctuation
Free Energy of Dimerization in water and 50% v/v water-methanol solvents
S
-4---
ý--- - -- - --rI
I--II g
I
-
--- - I - - -4I
--I--- --I t--
I
- I
4.5
4
E
--
-- I-----SI
+--------I---
. . . . . .
3,5
I
Ii
.
I
i
I-
-
.
.
I
I
I
I
I
I
.
I----r4--------
--
---
.
.
-- T------ I
-----I
-
-
I --
-
-
I-
..
.
i
3.
i
2ý5
I
I
I
2-
3
3.5
4
4.5
5
5.5
6
7
6,5
7.5
8
8.5
Reaction Coordinate({dl+d2)/2) (Angstrom)
- water+methanol (50% v/v) -
water
Figure 5-4: Free Energy diagram of dimerization in water and 50% v/v watermethanol solvents.
energies, it is concluded that the dimer structure is more stable in 50% v/v watermethanol mixtures than in pure water solutions. The fact that the solubility of glycine
is much lower in water-methanol solvents than in water provides further evidence
for the higher stability of dimer structure in water-methanol solutions. This result
seemingly condradicts the hypothesis of existence of more monomer-like structures
in water-methanol solutions. The solution behavior is further explored by analyzing
the fraction of glycine molecules that exist as monomers and dimers in solutions of
glycine in water and 50% v/v water-methanol solutions at various supersaturations.
5.2
Solution Simulations of Glycine
Solutions of Glycine in water and 50% v/v water-methanol mixtures at various supersaturation ratios were simulated in NAMD. Supersaturatio ratio is defined as
(C - C)/C,, where C is the concentration of glycine in the solution and C, is the solubility limit in the solution. Concentrations are expressed as mole ratios, i.e. number
of moles of solute devided by the number of moles of solvent. Solubility of Glycine
in water is 0.059 mole ratio (0.059 moles of glycine in one mole of water) and in 5(.1%
v/v water-methanol mixture is 0.0088 mole ratio (0.0088 moles of glycine in 0.6912
moles of water and 0.3088 moles of methanol). Show below in Tables 5.1 and 5.2
are the various supersaturations and number of molecules of each component in the
solutions used in the simulations.
Table 5.1: System Details for glycine-water solutions.
Concentration
(C - C,)/C,
Number
of Number
Glycine
Water
0.0297
-0.5
100
3367
0.0445
-0.25
200
4489
0.0535
-0.1
200
3741
0.0653
+0.1
200
3061
0.0891
+0.5
300
3367
0.1188
+1
400
3367
of
These systems were simulated for four nanoseconds and the trajectories of the
last nanosecond were analyzed to find the extent of aggregation and the fraction of
molecules that exists as monomers and dimers in the solution. Geometric criterion
for defining a hydrogen bond is thoroughly tested for the solutions simulated. The
crystal structure of a-glycine consists of a centrosymmetric dimer unit (Figure 5-1)
with a Hg, ... OgI, distance of 2.1 Aand an Ng.--H, ... Og y angle of 1540. A cut-off
of 2.2 A and 1400 was found to be an optimal definition for hydrogen bond between
glycine. This same definition was used in a previous study by Hughes et al. [85j] :br
hydrogen bonding between glycine zwitterions in water. Figures 5-5 and 5-6 show t;he
variation of the fraction of glycine monomers with time in water and water-methanol
mixtures respectively.
Table 5.2: System Details for glycine-water-methanol
solutions.
-Concentration
Number
Number
Number
(C- Cs)/cs
Water
Methanol
Glycine
0.0044
-0.5
7855
3509
C1.0066
-0.25
7855
3509
75
0.0079
-0.1
75
6545
2924
75
0.0097
0.1
75
5355
2393
0.0132
+0.5
100
5236
2339
~"'-~---------0.0176
150
5891
2623
--~~···_I
Fraction of Monomers in the glycine water solutions
-ss(+1)
-
ss(+0.1) --
s(-0.5) ... avg(+1) ......... vg(+O. 1) --- avg(-0.5)
Time (ns)
Figure 5-5: Fraction of glycine molecules that exist as monomers in glycine water
solutions at three different concentrations is shown. The thick lines correspond to the
av4erage fraction of monomers.
Figure 5-7 shows the fraction of glycine molecules that exist as monomers in water
and water-methanol mixtures at various supersaturations studied.
Clearly as the
supersaturation is increased the number of monomers decreases in the both solvents.
Interestingly, we observed a higher fraction of monomers in 50% v/v water-methanol
mixture than in water. This result supports the hypothesis of the presence of higher
percentage of monomers in water-methanol mixtures and might seem inconsistent
wvitth the free energy calculations which predict a higher stability of dimers in the
Fraction of Monomers in the glycine water-methanol(50 % v/v) solutions
OA
0.4
0
0
0.1
03
31
32
3.3
3.4
3.5
3.6
37
3.8
39
4
Time (ns)
Figure 5-6: Fraction of glycine molecules that exist as monomers in glycine watermethanol solutions at three different concentrations is shown. The thick lines cor:espond to the average fraction of monomers.
presence of methanol additive. A simple equilibrium model is used with reasonable
assumptions to show qualitatively why higher monomer concentration can be expected
in water-methanol mixtures.
The dimerization reaction that occurs in the solution can be written as:
2 GLYmonomer
==~- Glycatamer 4==> Glydimer
The free energy diagram (Figure 5-4) is used to estimate the equilibrium constanits
for the reactions involved. The mass action law is used to caculate the approxirn;~te
monomer fraction in the solutions.
kT(-A
q)
Kleq
K
2
-
acatamers
kT
eq = e(
amonomers
-G2e
--
kT
adi
acatamers
Total number of glycine molecules can be expressed as,
Ngly
Ngly,monomer +
2
Ngly,catamer +
Ngly,monomer +
2
2
Ngly,dimer +
Ngly,catamer +
i(Ngly,ithaggregate)
2
Ngly,dimer
(5.1)
For simplicity, we assume the number of glycine molecules present in higher ag-
gregates to be negligible. This assumption obviously breaks down at higher concentrations of glycine. The activities of monomer and dimer species are approximated to
mo:le fractions which is a reasonable assumption at lower concentrations of glycine.
Kleq
=
K 2 eq
-
(5.2)
Xcatamer 2
Xmonomer
Xdimers
Xcatamer
(5.3)
The mole fractions can be expressed as a function of supersaturation ratio 'S'. If
the solubility limit of glycine in the solvent is c8, the number of solvent molecules
re(quired to obtain a supersaturation ratio of S is
Ngiy
(5.4)
(5.4)
Nsol ent = c(1
Xdimer
Xcatamer
onomer
(Ngly,dimer)
(55)
(N0 yc.ctamer)
(5.6)
(Nsolvent)
(Nsovent)
(Ngly,monomer)
(Ns(5.7)vet)
In Equations 5.5, 5.6 and 5.7 the number of glycine are assumed to much smaller
than the number of solvent molecules. Solving equations 5.1 to 5.7, we obtain
fly,moom,)))
Kleq(1 + K 2 eq) = 2(c(1(1 - S((fgly,monomer)
2
2(c,(1 + S))((fjy,monomer)2
(5.8)
The above equation is solved for fgty,monomer in both water and water-methanol
mixtures and is plotted against the supersaturation ratio (Figure 5-8). AGleq and
.AG2eg are estimated using the dimerization free energy plots. As can be observed,
the fraction of monomers is higher in the solvent with lower solubility and higher
dimer stability. If the solubility is decreased and AG for dimerization reaction is kept
con[stant, the monomer fraction at the same supersaturation increases. Similarly if
the AG for dimerization reaction is slightly increased (made more negative) without
changing the rest of the variables, the monomer fraction decreases. Therefore, at
similar supersaturation levels both the solubility limit and free energy of dimerizatih)n
play very important and opposing roles in governing monomer fraction of glycine in
solutions. We speculate that the increase in monomer fraction in water-methanol
mixtures stems from the fact that decrease in solubility is much larger than the
corresponding increase in dimer stability.
Regarding the applicability of the equilibrium thermodynamics in this scenario,
the actiavtion barrier for dimerization reaction is very low compared to thermal fluctuations and the diffusion of glycine in the solutions is reasonably high to reach
quasi-equilibrium with respect to the dimerization reaction even though the solutions
are supersaturated. But it is very important to note that the above equilibrium model
gives only a qualitative picture beacuse of the various assumptions used which are
valid only at very low concentrations of glycine. The large difference in calculated
and predicted monomer fractions from the model should also be attributed to the
presence of a major fraction of glycine molecules in higher order aggregates.
Shown in Figure 5-9 is the fraction of glycine molecules that exist as dimers
(catamers & centrosymmetric dimers) in the solution. The plots show a relatively
small number of glycine dimers in the solutions. The water-methanol solvents have a
higher fraction of dimers at very low supersaturations but at higher supersaturation
ratios the difference in dimer fractions is very small. The fraction of dimers in the'
solution does not increase with concentration of glycine. This unexpected behavior
could be attributed to the presence of higher order aggregates where the interactions
between the monomers, dimers and the aggregates also has to be considered. Future
studies will focus on such interactions.
Fraction of glycine that exist as monomers in the solution
---- water -e- wat±methano I
0.6 E-·
0.5
a)
2o
•0.4
0
t3
0*
Z0.3
r=4
0.2
. .....
...
..............
.........
..
0.1 -..-
-0.2
-0.4
0
0.2
0.6
0.4
0.8
1
1.2
Supersaturation Ratio (c-cs)/cs
Figure 5-7: Fraction of glycine molecules that exist as monomers in water and watermethanol mixtures at various supersaturations.
Fraction of Monomers as a function of supersaturation ratio
0
W..
..
.... ....... .. .... ..... .. ...
-5
--------------
0.2
...
0.2
-
W.+-
-n
=
,c.
MOA A r1_1
.
n rz L. -1
1)2~_n
1-, 11,
,
=-
-
ca
50 % v/v water-methanol c,=0.0088,AG1=-l.1 kcal/mol, AG2=-1.0 kcal/mol
--- Hypothetical c,=0.0088.,AG1=-1.5 kcal/mol,AG2=-1.5 kcal/mol
/mo,
=-
.6
c
o
I
0
I
0.5
Supersaturation Ratio (S=(C- C)/CI)
Figure 5-8: Predicted fraction of glycine monomers in water and water-methanol
solvents and a hypothetical solvent at various supersaturations.
60
Fraction of glycine molecules that exist as dimers in the solution.
I-*-water --+-wat+methanol
0.6
0.5
....... ...........
..........
........................ . .......
... ........... ................. ......
0.4
I
--
0.2
0.1
L,
-'.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
Supersaturation Ratio (c-cs)/cs
Figure 5-9: Fraction of glycine molecules that exist as dimers in the solution of water
and water-methanol solvents.
5.3
Simulations of a & / Glycine Interfaces
In this section MD Simulations of glycine crystal interfaces were performed to explore
the solvent effect on crystal growth. Molecular dynamics and Free Energy calculations
were performed in the previous section to understand the increase in glycine monomers
in the presence of methanol additive. It was suggested that these monomers dock onl.to
the (010) interface a-glycine and further inhibit crystal growth. Simulations we,.re
carried out to explore the interface behavior of a-glycine with and without monomer
units at the interface. Various features like density, specific adsorption sites, energetics
etc. are evaluated to obtain insights in the solvent effects on the crystal interface.
The results are discussed in the following sections.
5.3.1
System Setup
The simulation cells consisted of a and 3 glycine crystals with (010) and (010) faces
exposed to water and water-methanol solutions.In the case of a-glycine two cas•es
were considered, one with C-H bonds exposed and the other with N-H bonds
exposed at the interface. Figures 5-10 and 5-11 show snapshots of the a-glycine
interfaces and the 0-glycine interfaces are shown in Figure 5-12. Water is the solvent
used when the interface consists of exposed C-H bonds in a-glycine and watermethanol solution is used for exposed N-H bonds in a-glycine and 3-glycine. This
is because ,-glycine is only crystallized in water-methanol solutions and an interface
with exposed N-H bonds in a-glycine is only possible when methanol is added to
water. The concentration of methanol is 0.3088 mole fraction (corresponding to 50%
v/v water-methanol mixture).
Periodic boundary conditions were applied in all three dimensions. When simulating such interfaces it is important to have the thickness of the solvent layer to
be at least four times the cutoff radius and the thickness of the crystal slab to be
at least two times the cutoff radius to prevent any artifacts caused due to the use
three dimensional periodic boundary conditions, such as interactions between water
molecules at one crystal interface with the water molecules at the other face (across
the. crystal boundary). The cutoff radius used was 14
A. The
thickness of the crystal
slalb is 48 A in a-glycine (corresponding to 9X4X9 crystal unit cells) and the thickness of the water layer is about 80
A, both well
within the constraints. The OPLS
force field with modified a-hydrogen parameters validated in the earlier section was
used for glycine, and OPLS parameters were used for water and methanol. Ewald
summation is used to calculate the electrostatic interactions for the system.
Figure 5-10: Schematic of the interface with water and (010) face of a-glycine with
C- -H groups exposed.
Figure 5-11: Schematic of the interface with water-methanol mixture and (010) fiace
of a-glycine with N-H groups exposed
Figure 5-12: Schematic of the interface with water-methanol mixture and (a) (010)
and (b) (010) of 3-glycine.
5.3.2
Density Profiles
Comparing the density of the solvent on different interfaces provides qualitative description of the extent of solvent adsorption. For the calculation of density, the system
is divided into bins of 0.25 Aperpendicular to the interface and the center of mass of
solvent molecules is used. The center of mass of glycine molecules at the interface
is used to define the crystal-solvent interface (y = 0). Density plots on both (010)
and (010) interfaces are shown in the figures below. The peaks in the density profiles
close to the interface indicate distinct surface layers present on the interface. Moving
away from the surfaces, the density becomes more uniform. The width of these peaks
is used to define the structured layers of water. Figure 5-13 shows the density profiles
of water on the (010) and (010) interfaces when the C-H groups are exposed (water
is the only solvent). The density profiles are symmetrical due to the symmetry of the
molecules in the unit cell at (010) and (010) faces.
Density Distribution of Water on alpha {010} and (0-10} interfaces with exposed C--H
groups
1,4
-
---------- ---------~-----------c------------------
'V
0,L2
0-2
1
0
-
-----------
-
1
2
3
4
----- ---
----------
5
6
------
7
8
9
10
11
12
Distance from interface (Angstrom)
--••
010) --'0-10
Figi.!ure 5-13: Density of water on (010) and (010) interfaces of a-glycine with C-H
grloups exposed at the interface.
Figure 5-14 describes the density profiles of water on (010) and (010) interfaces
when the N-H groups are exposed. As can be seen from the plots (010) has slightly
higher water adsorption than the (010) interface.
Similar plots for methanol are
sh:lwn in Figure 5-15. Since we are interested in growth inhibition by the solvent
on the faster growing interface, the (010) interface is used for analysis. Figure 5-16
sh<:'ws the difference in water density profiles and the extent of adsorption when the
C--H groups are exposed and when the N-H groups are exposed at the interface. It
ca.i be clearly observed that water penetration is more in the case of exposed N-H
64
Density Distribution of water on alpha (010) and (0-10) interfaces with exposed N--H
gropus
2
1.8 -
I------------------- -----------------------------------------------
---------------
1.6 -
-----------
--- -
---- ---------- -----4-------------------------------~------I----------
-----------
-----
----L------------------------------·-------L-------------------------
1.4 1.2 -
--------------- -- 0.8 -
+~ -- ----------------------------r---------------------
:- Q
- ---------I------------ - ------
---------------- -------_~_~~~~~
0.6 0.4 -
--------
:--------------.~5---- -----1----- .. ;..------------------c-----------------------~-----
--------------------
0.2 -
1 ·-
0O
0
1
2
3
4
5
6
7
8
9
10
11
12
Distance from the interface (Angstrom)
I-
{010} --(0-10}
Figure 5-14: Density of water on (010) and (010) interfaces of a-glycine with N- H
groups exposed at the interface.
Density Distribution of Methanol on alpha (010) and {0-10} interfaces with exposed N--H
groups
r------~----------------------
-- -------------------p-----------------------------------------------
0.5 -
----------i----- -----------------------
---
----------0.4 -
0,4
c 0.3 -
0.2 ------------·--
-- ----- ---------------------------------- ------
-------------------
0.1-
0C
0
1
2
3
4
5
6
7
8
9
10
11
1,
Distance from the interface (Angstrom)
I-
{0oo10} -- {0-10
I
Figure 5-15: Density of methanol on (010) and (010) interfaces of a-glycine with
N-H groups exposed at the interface.
65
Density profile on alpha (010} interface with C--H and N--H groups exposed
"` -`'~
~ ^~'~~''~"~' ""'~~'~~~-
2 -
~~'
--------------
i'
1.8 -
1.4 -
......
.....
i.......
..........
i.....
.....
i.....
,........
i..........
•..•.......
......
-------r--------~----w.~.
------- -- l\~----- ----- i------------- -----------------------
----
1,2-
------------------------------
--------
. . ~---------r-l-----......i .... l~- .... ..
.......i....
0.8
0.6
-- -------
ii
...•
i
7
8
--- -------
---------------i
0A4
0.2
0
0
1
2
3
4
5
6
9
10
11
12
Distance from the interface
-4-C--H exposed (water) ---- N-H exposed (water) --
N--H exposed (methanol)
Figure 5-16: Density comparison of solvent molecules on (010) interface of a-glycine
wit;h C-H and N-H groups exposed at the interface.
gro:,ups. In both cases two water layers are observed but the layers are closer to the
interface with narrower widths in the case of exposed N-H groups implying more
ad4sorption as predicted by the hypothesis. An important observation made possible
through molecular dynamics is the behavior of methanol at the interface. There is
only one peak of methanol observed which is present after the peak of second layer
of 1water implying that water plays the major role in inhibiting growth.
Density plots for 3-glycine interfaces are also plotted in Figures 5-17 and 5-18.
As previously reported [4], the (010) interface of 0-glycine grows at a slower rate
than the (010) face. Although the (010) face is very similar to (010) face of ac-glycine
vwith C-H groups exposed and the (010) face is very similar to (010) face of a-glycine
wit;h N-H groups exposed, the density profiles look different. In the case of 3-glycine
both water and methanol solvents show varying density profiles at (010) and (010)
int.erfaces and hence both the solvents play a role in inhibiting crystal growth.
66
Density Distribution of Water at beta {010) and (0-10} interfaces
1.8 -
1.6 -
..
- ---- ------ I
- ----------- ~/------.~
L --------------
------ -----
---------
-----------
1.4 ----------
:-----------\-------:---------- ------------------------------------------
1.2 -
-
----------- -----
1
-----
•------- - -...
-- --S0.8 ...........0.6 -
l---------------
0.4 -
·---------
- - .--------- - ---- --.-- - - -- .........- - - ------------------- ---------
----------- - -
~---- -------------;
---------------------L-----
---------- ------
0.2
---------------------------------------------.----.
----------
------ -----
-----------
----------:------------------
----------
----
I-------------------- -------
0
0
1
2
3
4
5
6
7
8
9
10
11
L2
Distance from the interface (Angstrom)
(--010}
-U-O-101
Figure 5-17: Density of water on (010) and (010) interfaces of 3-glycine.
Density of Methanol on beta (010) and (0-10} interfaces
0.90.80.7
--------------------------------------------
-------- ------------------ ----------------------
0.6 - 0.5 -
0.4 -
;
j ----------------------------
--------------0.3 0.2 -
----- ------- ----- ----
---------------------- -----------------------
-------------------- -------- ----- ---------- ---------------- --------------- -------
0.1
01
0
1
2
3
4
5
6
7
8
9
10
11
j4
Distance from interface (Angstrom)
--- (0io0} --
(0-10,
Figure 5-18: Density of methanol on (010) and (010) interfaces of a-glycine
5.3.3
Scatter Plots & Energetics
To obtain a quantitative picture of the structured layers close to the interface with
C-H and N-H groups exposed, the center of mass motion of solvent molecules
is explored for 50 picoseconds during the simulation. This is done by plotting the
67
trajectories of center of mass of solvent molecules in the layers defined by the density
plots. The widths of the first two peaks are used and the scatter plots were plotted
for the two layers with water and water-methanol solvents.
Figure 5-21 and Figure 5-22 show the scatter plot of solvent in the first structured
layer. As can be observed the first layer for water-methanol solvent is less populated
but more concentrated at specific sites on the interface. The first layer for water
solvent also exhibits preferential absorption on specific sites although the order is less
compared to the case with water-methanol solvent. These adsorption sites on the interfaces with C-H bonds and N-H bonds exposed are identified from the snapshots
of these layers taken from the simulations. Figure 5-19 illustrates the structured layer
observed at the interface with C-H bonds exposed. The water molecules form strong
Ha.,ater"" COO- hydrogen bonds and weak O
te
- - - CH hydrogen bonds. Similar
observations are made for the structured water layer on the interface with N-H bonds
exposed (Figure 5-20). The water molecules in this case form strong Hwater ... COOand Owater- - NH+ hydrogen bonds.
These specific islands of adsorption are further studied through interaction energy
calculations. Energetic calculations similar to the ones performed by Anwar et al.
[711 were performed. The interaction energy specified in Table 5.3 is the electrostatic
and Lennard-Jones interactions between one water molecule and the entire crystal
avve:raged over time. This energy is used to categorize the strength of binding for the
solvent molecules on the interface. An increase in interaction energy of about 36.2
K. 1/mol is observed when the N-H bonds are exposed at the interface.
Figure 5-19: Snapshot of the first structured layer on a-glycine interface with C-- H
groups exposed.
I;-
Figure 5-20: Snapshot of the first structured layer on a-glycine interface with N-- H
groups exposed.
Table 5.3: Interaction Energy of water at the specific sites with the a-glycine interface
Interface
Interaction Energy at specific sites
(KJ/mol)
(010) a-glycine with C-H bonds exposed -33.7
(010) a-glycine with N-H bonds exposed -69.4
Similar observations are made for the second layer with water and water-methanol
solvents (Figures 5-23 and 5-24), there is no preferential adsorption in the second layer
but the order is greatly reduced when the C-H groups are exposed at the interfac:e.
This is a clear indication of increased interactions for the solvent with N-H groups
than with the C-H groups.
projection of water with C--H exposed inthe first layer (50 ps)
-..-~Center
. _I·~-..l-iof Mass._I^·
i. . _·_-11 ......^ ^i·.(.··-.···· ^ 1_.·.·11.1
_I1 _~i_
-CF4,1
I---
o
I
I
.X
I
i
I
-5
-10
~--~I
0 ..•.
I
___~ __
Di•.0 0 A4 5)
X - Distance (Angstrom)
10
IS
2
22
Figure 5-21: Center of Mass distribution of water molecules on the a-glycine interface
with C-H groups exposed in the first structured layer.
Il
Center of Mass projection
of ---waterwih
N-H exposed inthe first layer (50 ps)
----------- - ~-----------------
- ·-------------·------
!
T~~-~
t
|I
i
I
-~~
I
I
II
I
I
itI
-
~"-·rsnrr~·
~T
.
.
I
I
I
I
I
I
i
I
i
I
I
I
S-+- (;I
--
I
~
.
...--
i
-~-t1
-- -
±.,..
9vwd,,1
,,
I
I
I
I
I41
IC
I
I I
I~~.#
I
-
I
I
I
I
'C"'
- -II
rl
-r+
I
I
.20
I
II
t
I
.25
4
1
|~-----!
$
i
-4-.-1.
i
I
I
I
II
-AUI
.
.
L.
.....
n
...
.
.
I
.. •
,|
|
X - Distance (Angstrom)
I
I
to
I
I
J
Figure 5-22: Center of Mass distribution of water molecules on the a-glycine interfaice
with N-H groups exposed in the first structured layer.
Center of Mass p~ojectlon of waler with C-4H exposed inthe second layer (50 ps)
I
I
I
F
i
I
I
I
I
I
I
i
I
F
I
I
21
2r ,,,,
arl-rr*l
,---
-LII
----
rn~r-
111-11
C
---
-4
----
.10,
--
-11
-·
irr-rs
-3S
30
"""·~(
1-
Z
.20
-15
-10
~-~--------~-
.. ................................
..... ...........
...... .. .............L.......................
• ...................
....J
4
0
5
10
15
X - Distance (Angstrom)
..... . .L•
5
~1
X
Figure 5-23: Center of Mass distribution of water molecules on the a-glycine interface
with C-H groups exposed in the second structured layer.
Center of Mass projection of water and methaol with N-H exposed inthe second ayer 00 ps)
A·
i
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
SI
a
IF
----
----
--
I
-40
. ...
.. .
I
`.•.
-
-25
I
I
1
.
-20
.
-45
I
I
-10
~--
I
I
Ii
I
.
I
.a|
I
-5
0
5
X - Distance (Apgstrom)
I
I
J
Figure 5-24: Center of Mass distribution of water and methanol molecules on the
a-glycine interface with N-H groups exposed in the second structured layer.
Chapter 6
Conclusions & Future Work
The force fields and parameters available in literature have been tested for their
ability to simulate all the glycine polymorphs. AMBER failed to reproduce the a
and 3 glycine structures and CHARMM failed in simulating a and 7 glycine. The
OPLS force field and parameters were successful in maintaining the hydrogen bond
networks of glycine polymorphs but showed large deviations in lattice parameters.
Better agreement with experimental observations was obtained by slightly modifying
the Lennard-Jones parameters of the hydrogen atoms attached to the a-carbon. The
Lennard Jones parameters for this hydrogen atom were adapted from AMBER parameters for glycine zwitterion. These parameters showed considerable improvement
both in lattice parameters and hydrogen network stability. The parameters were
further tested for stability in molecular dynamics. Constant pressure (1 atm) and
temperature (300 K) simulations of the polymorphs for one nanosecond with triclinic
boundary conditions were performed. The lattice parameters were found to be stable
and within reasonable limits of the experimental values. The force field and parameters were also successfully vailidated for their ability to model solution behavior in
water by comparing coordination numbers with experimental observations.
The solution behavior of glycine has been studied to investigate the reasons behind crystallization of 3-glycine in water-methanol mixtures. Weissbuch et al. [4j
postulated an increase in the glycine monomer units in the presence of the methanol
additive, but the reason was unclear. Free energy calculations for the dimerization
reaction were performed with and without a methanol additive. The reactant state
corresponds to two glycine monomers. The product state is the double hydrogen
bon::ded dimer unit. The free energy difference and the activation barrier for the
reaction (Figure 5-4) indicate higher stability of the centrosymmetric dimer unit in
50WI v/v water-methanol mixtures. Glycine dimer with one hydrogen bond was found
to be an intermediate state. Solutions of glycine in water and water-methanol mixtures were simulated over various supersaturations and the fraction of monomers
and dimers were calculated. Over the supersaturation levels explored a higher fraction of monomers was observed in water-methanol mixtures. The higher fraction of
mo.nomers was interpreted using simple mass action laws and equilibrium thermodynarmics. It was concluded that at similar supersaturation levels solubility and dimer
stability play opposing roles in governing the monomer fraction in solutions and the
drastic decrease in solubility due to the presence of methanol additive results in an
inc:reased fraction of monomers.
The nuclei of a-glycine type that are formed in the solutions are inhibited at
the crystal growth stage. An increase in glycine monomer units will result in the
formation of interfaces, with N-H bonds exposed on the a-glycine interface. The
ability of solvent to inhibit the crystal growth at these interfaces is explored through
simulations of a-glycine interfaces with both C-H and N-H bonds exposed. The
density profiles, scatter plots, and energetics of interactions on the interface clearly
indicate higher adsorption of the solvent at the interface with exposed N-H bonds.
This would result in an increase in the barrier,
AGdeh
(Figure 2-4), for stripping the
solvent molecules from the interface and hence an inhibition of crystal growth. The
den:sity profiles clearly indicate that water plays a dominating role in inhibiting crystal
growth when growth occurs through docking of monomer (010) crystal interface which
expose N-H groups to the solvent.
It is concluded that the role of the methanol additive is to increase the percentage
of glycine monomer units in the solution. Methanol being less polar than water has
a much lower solubility of glycine. The stability of a-glycine growth unit increases in
504;,c v/v water-methanol mixtures. If only the growth synthon hypothesis were to be
considered, a-glycine would be predicted in water-alcohol solutions. Link hypothelsis
postulates that the most stable growth unit in the solution would present itself in the
final crystal structure. We conclude that both the stability of the growth synthon and
the supersaturation play important roles in determining the final polymorph outcomi:e,
and in this case the effect of supersaturation dominates the stability of the a-glycine
growth synthon, resulting in the growth inhibition of a-glycine polymorph.
The interface simulations of -y-glycine proved challenging due to the difficulty of
simulating a polar crystal interface. The y-glycine crystal has a net dipole mom,:nt
perpendicular to the {001} interface. Simulations of such interfaces have been carried
out in the literature but are limited to ionic crystals. Development of a method to
simulate polar interfaces in organic crystals would lead to key insights on the 'relay
growth' mechanism.
In summary, key qualitative insights into the methanol additive effect on polymorph selection during glycine crystallization were obtained through molecular dynamics simulations of solution behavior of glycine. But in order to obtain any quantitative results that would form the basis for a rational algorithm for the screening of
solvents, it is absolutely necessary to understand the nucleation mechanism and also
the structure of the critical nucleus. Future studies in this direction would facilitate
in the determination of rate equations and in isolation of the key properties of the
solvent that effect the polymorphic outcome.
Bibliography
[1] Lian Yu, G. A. Stephenson, C. A. Mitchell, C. A. Bunnell, S. V. Snore:k,
J. J. Bowyer, T. B. Borchardt, and J. G. Stowell adn S. R. Byrn. Therm]ochemistry and conformational polymorphism of a hexamorphic crystal system.
J. Am. Chem. Soc., 122(4):585-591, 2000.
[2] R. J. Davey, K. Allen, N. Blagden, W. I. Cross, H. F. Liebermann, M. J. Quayile
S. Righini, L. Seton, and G. J. T. Tiddy. Crystal engineering - nucleation, the
key step. CrystEngComm, 4(47):257-264, 2002.
[3] P. Bennema. Analysis of crystal growth models for slightly supersaturated solutions. Journal of Crystal Growth, 1:278-286, 1967.
[4] I. Weissbuch, V. Yu. Torbeev, L. Leiserowitz, and M. Lahav. Solvent effect on
crystal polymorphism: Why addition of methanol or ethanol to aqueous solutions
induces the precipitation of the least stable /3 form of glycine. J. Am. Chem. S;'.,
110:7690-7697, 2001.
[5] I. WVeissbuch, L. Addai, and L. Leiserowitz. Molecular Recognition at Crystal
Interfaces. Science, 253(5020):637-645, 1991.
[6] J. Kottalam and D. A. Case. Dynamics of ligand escape from the heme pocket
of myoglobin. J. Am. Chem. Soc., 110:7690-7697, 1988.
[7] J. W. Mullin. Crystallization. Butterworth-Heinemann, fourth edition, 9 M\lay
2001.
[8] H.N. de Armas, O.M. Peeters, N. Blaton, D.J.A. De Ridder, and H. Schenk. Xray powder diffraction data and crystal data of polymorphic form 2 of carnidazole. Powder Diffraction, 21:56, 2006.
19] J. Bauer, S. Spanton, R. Henry, J. Quick, W. Dziki, W. Porter, and J. Mw.)rris. Ritonavir: An extraordinary example of conformational polymorphism.
PharmaceuticalResearch, 18(6):859-866, 2001.
[10] T. L. Threlfall. Analysis of organic polymorphs. a review.
120(10):2435, 1995.
The Analtyst,
[11. H. Ando, M. Ishii, M. Kayano, and H. Ozawa. Effect of moisture on crystallization of theophylline in tablets. Drug Development and Industrialpharamacy,
18(4):453-467, 1992.
[121 K. R. Morris, A. W. Newman, D. E. Bugay, S. A. Ranadive, A. K. Singh,
M. Szyper and. S. A. Varia, H. G. Brittain, and T. M. Serajuddin. Characterization of humidity-dependent changes in crystal properties of a new hmg-coa
reductase inhibitor in support of its dosage form development. International
Journal of Pharmaceutics,108(3):195-206, 1994.
[131 M. Otsuka and Y. Matsuda. Effects of environmental temperature and compression energy on polymorphic transformation during tabletting: Pharmaceutical
compaction Research Laboratory and Information Center. Annual symposium.
Drug development and industrialpharmacy, 19(17-18):2241-2269, 1993.
[14] I. Weissbuch, M. Lahav, and L. Leiserowitz. Toward stereochemical control,
monitoring, and understanding of crystal nucleation. Crystal growth e design,
3(2):125-150, 2003.
11511 FJJ Leusen. Ab initio prediction of polymorphs. Journal of Crystal Growth,
166(1):900-903, 1996.
[161 A. Gavezzotti, G. Filippini, J. Kroon, BP Van Eijck, and P. Klewinghaus. The
Crystal Polymorphism of Tetrolic Acid (CH 3 CCCOOH): A Molecular Dynamics
study in solution, and a Crystal Structure Generation. Chemistry(Weinheim),
3(6):893-899, 1997.
[17] EB Treivus. Solvent effect on the kinetics of crystal growth. Russian Chemical
Reviews, 61(7):673-682, 1992.
[18i RJ Davey. The Role of the Solvent in Crystal Growth from Solution. J. Crystal
Growth, 76:637-644, 1986.
[191 A. C. Zettlemoyer. Nucleation Phenomena. Marcel Dekker, 1969.
[201 W. Ostwald. Z. Phys. Chem., 22:289, 1897.
[211 G. Nichols and C.S. Frampton. Physicochemical characterization of the orthorhombic polymorph of paracetamol crystallized from solution. Journal of
PharmaceuticalSciences, 87(6):684-693, 1998.
[221 Becker D. and Ddring W. The kinetic treatment of nuclear formation in supersaturated vapors. Annalen der Physik, 24:719-752, 1935.
[23P1M. Volmer and A. Weber. Keimbildung in iibers~ittigten gebilden.
Chem., page 227, 1925.
Z. Phys.
[241 M. Volmer. Kinetik den Phasenbildung. Leipzig, Dresden: Th. Steinkopf, 1939.
Vgl. hicrzu: a) Vetter, KJ: Electrochemical Kinetics, 1967.
[25] D. Turnbull and JC Fisher. Rate of Nucleation in Condensed Systems.
'T1he
Journal of Chemical Physics, 17:71, 2004.
126] D.W. Oxtoby and R. Evans. Nonclassical nucleation theory for the gas-liqiitid
transition. The Journal of Chemical Physics, 89:7521, 1988.
[27] D.W. Oxtoby. Nucleation of first-order phase transitions.
31(1):91, 1998.
Acc. Chem. [i'es,
[28] J.P. Behr. The Lock-and-key Principle: The State of the Art-100 Years on, pIa)ge
173. Wiley, 1994.
[29] A. Gavezzotti and G. Filippini. Self-organization of small organic molecules
in liquids, solutions and crystals: static and dynamic calculations.
Communications, 1998(3):287-294, 1998.
Chem&ical
[301 PR ten Wolde and D. Frenkel. Enhancement of protein crystal nucleation by
critical density fluctuations. Science, 277(5334):1975-8, 1997.
[31] J. Anwar and P.K. Boateng. Computer Simulation of Crystallization from So:lution. Springer Ser. Chem. Phys, page 43, 1987.
[32] J.D. Shore, D. Perchak, and Y. Shnidman. Simulations of the nucleation of Ag Br
from solution. The Journal of Chemical Physics, 113:6276, 2000.
[33] S. Chattopadhyay, D. Erdemir, J.M.B. Evans, J. Ilavsky, H. Amenitsch, C.U.
Segre, and A.S. Myerson. SAXS study of the nucleation of glycine crystals from
a supersaturated solution. Crystal Growth & Design, 5(2):523-527, 2005.
[34] R. Radhakrishnan and B.L. Trout. Nucleation of Hexagonal Ice (I h) in Liquid
Water. J. AM. CHEM. SOC, 125:7743, 2003.
[35] J.M. Leyssale, J. Delhommelle, and C. Millot. Atomistic simulation of the hoil:logeneous nucleation and of the growth of N crystallites. The Journal of Chemnical
Physics, 122:104510, 2005.
[36] J.M. Leyssale, J. Delhommelle, and C. Millot. Molecular simulation of the honvlogeneous crystal nucleation of carbon dioxide. The Journal of Chemical Phys:ics,
122:184518, 2005.
[37] S. Parveen, RJ Davey, G. Dent, and RG Pritchard.
Linking solution cheim-
istry to crystal nucleation: the case of tetrolic acid. Chemical Communicatio11s,
2005(12):1531-1533, 2005.
[38] L.R. MacGillivray and M.J. Zaworotko. Crystal and molecular structure of 2,
6-dihydroxybenzoic acid. Journal of Chemical Crystallography, 24(10):703-71)5,
1994.
139.1 M. Gdaniec, M. Gilski, and GS Denisov. -y-Resorcyclic acid, its monohydrate
and its pyridinium complex. Acta crystallographica.Section C, Crystal structure
communications, 50:1622-1626, 1994.
1401 RJ Davey, N. Blagden, S. Righini, H. Alison, MJ Quayle, and S. Fuller. Crystal
Polymorphism as a Probe for Molecular Self-Assembly during Nucleation from
Solutions: The Case of 2, 6-Dihydroxybenzoic Acid. Crystal Growth & Design,
1(1):59-65, 2001.
[411 R,.J. Davey, W. Liu, M.J. Quayle, and G.J.T. Tiddy. In Situ Monitoring of
Crystallization Processes Using Synchrotron X-ray Diffraction: The Search for
Structural Precursors. Crystal Growth & Design, 2(4):269-272, 2002.
[42.1 RJ Davey, G. Dent, RK Mughal, and S. Parveen. Concerning the Relationship
between Structural and Growth Synthons in Crystal Nucleation: Solution and
Crystal Chemistry of Carboxylic Acids As Revealed through IR Spectroscopy.
Crystal Growth & Design, 6(8):1788-1796, 2006.
1431 A. F. Wells. New Investigations in Crystallography and Crystal Chemistry:
Translated into Russian, volume 5. Izdatinlit, Moscow, second edition, 1950.
[441
D. H. Watson. In V. E. Cosslett, editor, Proc. of the Third International Conference on Electron Microscopy, page 497, London, 1954. Royal Microscopical
Society.
[45.1 W. Kleber and H. Raidt. Z. Phys. Chem., 222, 1963.
[46.1 R. J. Davey. Current Topics in Materials Science, volume 8, chapter 6. North
Holland, Amsterdam, 1982.
1471
P. Bennema and G. Gilmer. Crystal Growth: An Introduction,chapter 10. North
Holland, Amsterdam, 1973.
[4811 WK Burton, N. Cabrera, and FC Frank. The Growth of Crystals and the Equilibrium Structure of their Surfaces. Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences, 243(866):299-358,
1951.
1491 LJW Shimon, M. Vaida, L. Addadi, M. Lahav, and L. Leiserowitz. Molecular
recognition at the solid-solution interface: a relay mechanism for the effect of
solvent on crystal growth and dissolution. Journal of the American Chemical
Society, 112(17):6215-6220, 1990.
[501 LJW Shimon, FC Wireko, J. Wolf, I. Weissbuch, L. Addadi, Z. Berkovitchyeelin, M. Lahav, and L. Leiserowitz. Assignment of Absolute Structure of Polar
Crystals Using Tailor-Made Additives. Solvent Surface Interactions on the Polar
Crystals of a-Resorcinol,(R, S) Alanine and Y-Glycine. Molecular Crystals and
Liquid Crystals, 137(1):67-86, 1986.
[51] LJW Shimon, M. Lahav, and L. Leiserowitz. Design of stereoselective etchanIts
for organic crystals. Application for the sorting of enantiomorphs and dirict
assignment of absolute configuration of chiral molecules. Journalof the Americnan
Chemical Society, 107(11):3375-3377, 1985.
[52j E.V. Boldyreva, V.A. Drebushchak, T.N. Drebushchak, I.E. Paukov, Y.A. Kovalevskaya, and E.S. Shutova. Polymorphism of glycine: Thermodynamic aspects. Part I. Relative stability of the polymorphs. Journal of Thermal Analysis
and Calorimetry, 73(2):409-418, 2003.
[53] D. Bernal. Z. Krist., 78:363, 1931.
[54] J. Hengstenberg and F. V. Lenel. Z. Krist., 77:424, 1931.
[55] G. Albrecht and RB Corey. The crystal structure analysis of glycine. J. Amner.
Chem. Soc, 61:1087-1103, 1939.
[56j RE Marsh. A refinement of the crystal structure of glycine. Acta Crystallographica, 11(9):654-663, 1958.
[57] Fischer E. Ber. Deut. Chem. Ges., 38:2917, 1904.
[58] Y. Iitaka. The crystal structure of 3-glycine. Acta Crystallographica,13(1):35-45,
1960.
[59] Y. litaka. In Proc. of the Japan Academy, page 109, 1954.
[60] Y. Iitaka. The crystal structure of -y-glycine. Acta Crystallographica,11(3):225226, 1958.
[61] Y. Iitaka. The Crystal Structure of y-Glycine. Acta Cryst, 14:1, 1961.
[62] _. Kvick, WM Canning, TF Koetzle, and GJB Williams. An experimental stuldy
of the influence of temperature on a hydrogen-bonded system: the crystal strui.cture of y-glycine at 83 K and 298 K by neutron diffraction. Structural Crys;,allography and Crystal Chemistry, 36(1):115-120, 1980.
[63] WCM Lewis. The Crystallization, Denaturation and Flocculation of Proteins
with Special Reference to Albumin and Hemoglobin; together with an Appendlix
on the Physicochemical Behavior of Glycine. Chemical Reviews, 8(1):81-i16.5,
1931.
[64] D. Gidalevitz, R. Feidenhans'l, S. Matlis, D.M. Smilgies, MJ Christensen,
and L. Leiserowitz. MONITORING IN SITU GROWTH AND DISSOLUTION OF MOLECULAR CRYSTALS: TOWARDS DETERMINATION OF
THE GROWTH UNITS. Angewandte Chemie. International edition in English,
36(9):955-959, 1997.
[651 AS Myerson and P.I. Lo. Cluster formation and diffusion in supersaturated
binary and ternary amino acid solutions. Journal of crystal growth, 110(1-2):2633, 1991.
[66] J. W. Chew, S. N. Black, P. S. Chow, R. B. Tan, and K. J. Carpenter. Stable polymorphs: difficult to make and difficult to predict. CrystEnggComm,
9(2):128-130, 2007.
1671 C.S. Towler, R.J. Davey, R.W. Lancaster, and C.J. Price. Impact of Molecular Speciation on Crystal Nucleation in Polymorphic Systems: The Conundrum
of Glycine and Molecular'Self Poisoning'. Journal of the American Chemical
Society, 126(41):13347-13353, 2004.
[681 Z. Berkovitch-Yellin, L. Addadi, M. Idelson, L. Leiserowitz, and M. Lahav. Absolute configuration of chiral polar crystals. Nature, 296(5852):27-34, 1982.
[691 FC Wireko, LJW Shimon, F. Frolow, Z. Berkovitch-Yellin, M. Lahav, and L. Leiserowitz. Effect of solvent on the growth of organic crystals. 1. The riddle of.
alpha.-resorcinol. The Journal of Physical Chemistry, 91(2):472-481, 1987.
[701 J.L. Wang, L. Leiserowitz, and M. Lahav. A correlation between surface wettability and solvent effect on crystal growth. The Nn-octyl-D-gluconamide/methanol
system. The Journal of Physical Chemistry, 96(1):15-16, 1992.
17111 M. Hussain and J. Anwar. The Riddle of Resorcinol Crystal Growth Revisited:
Molecular Dynamics Simulations of R-Resorcinol Crystal-Water Interface. J.
Am. Chem. Soc, 121(37):8583-8591, 1999.
[72.1 I. Weissbuch, L. Leisorowitz, and M. Lahav. Tailor-made and charge-transfer
auxiliaries for the control of the crystal polymorphism of glycine. Advanced
materials(Weinheim), 6(12):952-956, 1994.
[73.1 P.W. Carter, A.C. Hillier, and MD Ward. Nanoscale Surface Topography and
Growth of Molecular Crystals: The Role of Anisotropic Intermolecular Bonding.
Journal of the American Chemical Society, 116(3):944-953, 1994.
[741 F.H. Allen. The Cambridge Structural Database: a quarter of a million crystal
structures and rising. logo, 58(1 Part 3):380-388.
175.1 J.C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa,
C. Chipot, R.D. Skeel, L. Kale, and K. Schulten. Scalable molecular dynamics with NAMD. J. Comput. Chem, 26(16):1781-1802, 2005.
[761 BR Brooks, RE Bruccoleri, BD Olafson, DJ States, S. Swaminathan, and
M. Karplus. CHARMM: a program for macromolecular energy, minimization,
and dynamics calculations. Journal of computational chemistry, 4(2):187-217,
1983.
[77] WL Jorgensen and J. Tirado-Rives. The OPLS potential functions for proteins.
Energy minimizations for crystals of cyclic peptides and crambin. Journal of the
American Chemical Society, 110(6):1657-1666, 1988.
[78] Y. Duan, C. Wu, S. Chowdhury, M.C. Lee, G. Xiong, W. Zhang, R. Yaing,
P. Cieplak, R. Luo, T. Lee, et al. A point-charge force field for molecular rmiechanics simulations of proteins based on condensed-phase quantum mechani:.al
calculations. Journal of Computational Chemistry, 24(16):1999-2012, 2003.
[79] PP Ewald. Evaluation of optical and electrostatic lattice potentials. Ann. PhI's,
64:253, 1921.
[80] D.L. Veenstra, D.M. Ferguson, and P.A. Kollman. How transferable are hydrýog:en
Journal of Computational
parameters in molecular mechanics calculations?
Chemistry, 13(8):971-978, 1992.
[81] CM Freeman, JW Andzelm, CS Ewig, J.R. Hill, and B. Delley. The structuLre
and energetics of glycine polymorphs based on first principles simulation using
density functional theory. Chemical Communications,1998(22):2455-2456, 19!0)8.
[82] V. Bisker-Leib and M.F. Doherty. Modeling Crystal Shape of Polar Organilic
Materials: Applications to Amino Acids. Crystal Growth 6 Design, 3(2):221237, 2003.
[83] Y. Kameda, H. Ebata, T. Usuki, O. Uemura, and M. Misawa. Hydration Stru.cture of Glycine Molecules in Concentrated Aqueous Solutions. Bulletin of the
Chemical Society of Japan, 67(12):3159-3164, 1994.
[84] K. Leung and S.B. Rempe. Ab initio molecular dynamics study of glycine
intramolecular proton transfer in water. The Journal of Chemical Phy~:ics,
122:184506, 2005.
[85] C.E. Hughes, S. Hamad, K.D.M. Harris, C.R.A. Catlow, and P.C. Griffitlhs.
A multi-technique approach for probing the evolution of structural properties
during crystallization of organic materials from solution. Faraday Discuss'ions,
136:71-89, 2007.